Comparative Analysis of Hybrid Deep Learning Models for Dam Inflow Prediction: LSTM-GRU, CNN-LSTM, Attention-LSTM, and Transformer Approaches

Safavi, Maryam; Khashei-Siuki, Abbas; Hashemi, Reza; Piri, Jamshid; Ehteram, Mohammad

doi:10.22077/jwhr.2025.9938.1185

Document Type : Research Paper

Authors

¹ Ph.D Student, Department of Water Engineering, University of Birjand, Birjand, Iran

² Professor, Department of Water Engineering, University of Birjand, Birjand, Iran

³ Associate Professor, Department of Water Engineering, University of Birjand, Birjand, Iran

⁴ Associate Professor, Department of Water Engineering, University of Zabol, Zabol, Iran

⁵ Postdoctoral, Department of Water Engineering, University of Semnan, Semnan, Iran

https://doi.org/10.22077/jwhr.2025.9938.1185

Abstract

This study offers the first comprehensive comparison among four hybrid deep learning architectures—LSTM-GRU, CNN-LSTM, Attention-LSTM, and Transformer—for multipurpose dam inflow forecasting under severe hydrological variability. The study employed a 14-year dataset (168 observations, 2010-2023) obtained from Jiroft Dam in Iran and framed with hydrological and operational parameters including precipitation, reservoir capacity, agricultural discharge, and turbine functions. The LSTM-GRU architecture yielded the best performance by attaining 0.873 R² and 29.73 m³/s root mean square error (RMSE) during the validation procedure and demonstrating the best balance among accuracy and generalizability. The model robustness was confirmed by advanced validation methods including Taylor diagrams, violin diagrams, and statistical testing (Kolmogorov-Smirnov, Ljung-Box, and Breusch-Pagan tests). Seasonal analysis revealed a seven times change in flow rates ranging across winter maxima of 391.5 m³/s and autumn minima of 56.2 m³/s. The models showed a widespread tendency to predict lower peak flows (percentage bias, PBIAS: -14.34% to -20.86%), suggesting the presence of operational safety buffers. Precipitation–agricultural interactions were identified as the key forecasting variable (importance = 0.999). The model provides real-time support for decision-making on reservoir management, flood protection, and potable water supply under changing environmental circumstances and provides a validated model for AI-accelerated water resource management.

Keywords

Main Subjects

Water resources management

References

Ahrari, A., Sharifi, A., & Haghighi, A. T. (2024). Anthropogenic vs. climatic drivers: Dissecting Lake desiccation on the Iranian plateau. Journal of environmental management, 368, 122103.

Alhussein, M., Aurangzeb, K., & Haider, S. I. (2020). Hybrid CNN-LSTM model for short-term individual household load forecasting. Ieee Access, 8, 180544-180557.

Damansabz, A., Khajeh, M., Piri, J., & Ghaffari-Moghaddam, M. (2025). A novel GWO-DE-LSTM hybrid model for predicting statin drug solubility in supercritical carbon dioxide: a comparative analysis with traditional machine learning approaches. Network Modeling Analysis in Health Informatics and Bioinformatics, 14(1), 1-16.

Deb, D., Arunachalam, V., & Raju, K. S. (2024). Daily reservoir inflow prediction using stacking ensemble of machine learning algorithms. Journal of hydroinformatics, 26(5), 972-997.

Farhadi, A., Zamanifar, A., Alipour, A., Taheri, A., & Asadolahi, M. (2025). A hybrid LSTM-GRU model for stock price prediction. Ieee Access.

Galassi, A., Lippi, M., & Torroni, P. (2020). Attention in natural language processing. IEEE transactions on neural networks and learning systems, 32(10), 4291-4308.

Granata, F., & Di Nunno, F. (2025). Pathways for hydrological resilience: Strategies for adaptation in a changing climate. Earth Systems and Environment, 1-29.

Jiang, D., & Wang, K. (2019). The role of satellite-based remote sensing in improving simulated streamflow: A review. Water, 11(8), 1615.

Keshtegar, B., Piri, J., & Kisi, O. (2016). A nonlinear mathematical modeling of daily pan evaporation based on conjugate gradient method. Computers and Electronics in Agriculture, 127, 120-130.

Kim, S., Choi, K., Choi, H.-S., Lee, B., & Yoon, S. (2022). Towards a rigorous evaluation of time-series anomaly detection. Proceedings of the AAAI conference on artificial intelligence,

Li, X., Yang, X., Wang, X., & Deng, C. (2024). Agree to disagree: Exploring partial semantic consistency against visual deviation for compositional zero-shot learning. IEEE Transactions on Cognitive and Developmental Systems, 16(4), 1433-1444.

Liang, E., Tang, H., Liu, Y., Liu, S., Wu, J., Pan, W., Shang, Y., & Yin, S. (2025). A global synthesis reveals the role of strategic hydropower planning in mitigating adverse impacts of reservoir flooding. Renewable and Sustainable Energy Reviews, 217, 115723.

Livieris, I. E., Pintelas, E., & Pintelas, P. (2020). A CNN–LSTM model for gold price time-series forecasting. Neural computing and applications, 32(23), 17351-17360.

Mienye, I. D., Swart, T. G., & Obaido, G. (2024). Recurrent neural networks: A comprehensive review of architectures, variants, and applications. Information, 15(9), 517.

Ortiz-Partida, J. P., Fernandez-Bou, A. S., Maskey, M., Rodríguez-Flores, J. M., Medellín-Azuara, J., Sandoval-Solis, S., Ermolieva, T., Kanavas, Z., Sahu, R. K., & Wada, Y. (2023). Hydro-economic modeling of water resources management challenges: Current applications and future directions. Water Economics and Policy, 9(01), 2340003.

Piri, J., & Kisi, O. (2024). Hybrid non-linear probabilistic model using Monte Carlo simulation and hybrid support vector regression for evaporation predictions. Hydrological Sciences Journal, 1-29.

Pokharel, S. (2025). Towards Advancing Streamflow and Peak Flow Prediction With Machine Learning: Identifying Infrastructure at Risk The University of Nebraska-Lincoln].

Qian, X., Wang, B., Chen, J., Fan, Y., Mo, R., Xu, C., Liu, W., Liu, J., & Zhong, P.-a. (2025). An explainable ensemble deep learning model for long-term streamflow forecasting under multiple uncertainties. Journal of Hydrology, 133968.

Rithani, M., Kumar, R. P., & Doss, S. (2023). A review on big data based on deep neural network approaches. Artificial Intelligence Review, 56(12), 14765-14801.

Sajjad, M., Khan, Z. A., Ullah, A., Hussain, T., Ullah, W., Lee, M. Y., & Baik, S. W. (2020). A novel CNN-GRU-based hybrid approach for short-term residential load forecasting. Ieee Access, 8, 143759-143768.

Smith, J. D., Koenig, L. E., Sleckman, M. J., Appling, A. P., Sadler, J. M., DePaul, V. T., & Szabo, Z. (2024). Predictive Understanding of Stream Salinization in a Developed Watershed Using Machine Learning. Environmental Science & Technology, 58(42), 18822-18833.

Suzauddola, M., Zhang, D., Zeb, A., Chen, J., Wei, L., & Rayhan, A. S. (2025). Advanced deep learning model for crop-specific and cross-crop pest identification. Expert Systems with Applications, 274, 126896.

Thrun, M. C., Gehlert, T., & Ultsch, A. (2020). Analyzing the fine structure of distributions. PloS one, 15(10), e0238835.

Ullah, K., Ahsan, M., Hasanat, S. M., Haris, M., Yousaf, H., Raza, S. F., Tandon, R., Abid, S., & Ullah, Z. (2024). Short-term load forecasting: A comprehensive review and simulation study with CNN-LSTM hybrids approach. Ieee Access.

Uppalapati, S., Paramasivam, P., Kilari, N., Chohan, J. S., Kanti, P. K., Vemanaboina, H., Dabelo, L. H., & Gupta, R. (2025). Precision biochar yield forecasting employing random forest and XGBoost with Taylor diagram visualization. Scientific Reports, 15(1), 7105.

Wang, W.-c., Gu, M., Hong, Y.-h., Hu, X.-x., Zang, H.-f., Chen, X.-n., & Jin, Y.-g. (2024). SMGformer: integrating STL and multi-head self-attention in deep learning model for multi-step runoff forecasting. Scientific Reports, 14(1), 23550.

Wang, X., Zhou, J., Ma, J., Luo, P., Fu, X., Feng, X., Zhang, X., Jia, Z., Wang, X., & Huang, X. (2024). Evaluation and comparison of reanalysis data for runoff simulation in the data-scarce watersheds of alpine regions. Remote Sensing, 16(5), 751.

Zhang, D.-D., & Xu, J. (2024). Long-term monitoring of surface water dynamics and analysis of its driving mechanism: A case study of the Yangtze River Basin. Water, 16(5), 677.

Water Harvesting Research

Comparative Analysis of Hybrid Deep Learning Models for Dam Inflow Prediction: LSTM-GRU, CNN-LSTM, Attention-LSTM, and Transformer Approaches

References

References

Volume 8, Issue 2
September 2025
Pages 187-206

Comparative Analysis of Hybrid Deep Learning Models for Dam Inflow Prediction: LSTM-GRU, CNN-LSTM, Attention-LSTM, and Transformer Approaches

References

References

Volume 8, Issue 2September 2025Pages 187-206

Volume 8, Issue 2
September 2025
Pages 187-206