Çok Değişkenli Baraj Doluluk Oranı Tahmini için Kalıntı Tabanlı Hibrit BiLSTM–XGBoost Modeli: İstanbul Örneği

Year 2026, Volume: 14 Issue: 2 , 381 - 395 , 19.04.2026

Hikmet Canlı

https://doi.org/10.29130/dubited.1809519

https://izlik.org/JA83LJ89YL

Abstract

Küresel ısınma ve iklim değişikliği, dünya genelinde su kaynaklarını tehdit ederek sürdürülebilir su yönetimini zorlaştırmaktadır. Barajlar, özellikle İstanbul gibi büyük şehirlerde su temini açısından kritik öneme sahiptir. Bu çalışmada, İstanbul Büyükşehir Belediyesi Açık Veri Portalı’ndan elde edilen farklı veri setleri, günlük yağış, su tüketimi ve baraj doluluk oranları ile birleştirilerek baraj doluluk oranlarının tahmin edilmesi amaçlanmıştır. LSTM, Bi-LSTM, XGBoost ve Prophet gibi zaman serisi modelleri kullanılarak yapılan tahminlerin doğruluğu RMSE, MAE ve R² metrikleri ile değerlendirilmiştir. Sonuçlar, LSTM ve Bi-LSTM modellerinin en düşük hata oranlarıyla en başarılı performansı sergilediğini, Prophet modelinin ise en düşük doğruluğa sahip olduğunu göstermiştir. Elde edilen bulgular, derin öğrenmeye dayalı modellerin baraj su seviyesi tahmini için daha etkili bir yöntem olduğunu vurgulamakta ve bu tür modellerin sürdürülebilir su yönetimi açısından önemini ortaya koymaktadır.

Keywords

Sürdürülebilir su yönetimi , Zaman serisi modelleri , Derin öğrenme

Ethical Statement

Bu çalışmanın hazırlık sürecinde bilimsel ve etik ilkelere uyulduğu ve yararlanılan tüm çalışmaların kaynakçada belirtildiği beyan edilir.

A Residual-Based Hybrid BiLSTM–XGBoost Model for Multivariate Dam Fill Rate Forecasting: The Case of Istanbul

https://izlik.org/JA83LJ89YL

Abstract

Global climate change and increasing water demand pose serious challenges for sustainable water resource management, particularly in large metropolitan areas such as Istanbul. Accurate forecasting of dam fill rates is therefore critical for proactive water management and drought mitigation strategies. In this study, we propose a Residual-Based Hybrid BiLSTM–XGBoost (RBH-BiLSTM-XGB) model to forecast dam fill rates using multivariate time series data, including daily precipitation, water consumption, and historical reservoir levels obtained from the Istanbul Metropolitan Municipality Open Data Portal. The proposed approach combines the temporal dependency learning capability of Bidirectional Long Short-Term Memory (BiLSTM) networks with the residual error modeling strength of XGBoost, enabling the correction of systematic prediction errors produced by deep learning models. The proposed hybrid model is evaluated against standalone LSTM, BiLSTM, XGBoost, and Prophet models under different training–testing split scenarios (70–30%, 80–20%, and 90–10%). Model performance is assessed using RMSE, MAE, MAPE, and R² metrics. Experimental results demonstrate that the RBH-BiLSTM-XGB model consistently outperforms all benchmark models, achieving the lowest RMSE (0.0059), MAPE (0.66–0.86%), and the highest explanatory power (R² ≈ 0.9994). While deep learning models effectively capture long-term temporal dependencies, tree-based models such as XGBoost are shown to be effective in learning residual structures that deep networks fail to model. In contrast, the Prophet model exhibits poor performance due to its additive structure and limited capacity to represent complex multivariate interactions. The findings highlight the effectiveness of residual-based hybrid modeling for dam fill rate forecasting and demonstrate the potential of integrating deep learning and ensemble learning approaches to support data-driven and sustainable water resource management.

Keywords

Deep Learning , Sustainable water management , Time series models

Ethical Statement

This study does not involve human or animal participants. All procedures followed scientific and ethical principles, and all referenced studies are appropriately cited.

Supporting Institution

This research received no external funding.

Thanks

The author do not wish to acknowledge any individual or institution.

References

• Brownlee, J. (2017). Machine learning mastery with XGBoost and Scikit-Learn. Machine Learning Mastery.
• Bulut, M. (2021). Hydroelectric generation forecasting with long short term memory (LSTM) based deep learning model for Turkey. arXiv. https://doi.org/10.48550/arXiv.2109.09013
• Canlı, H., & Toklu, S. (2021). Deep learning-based mobile application design for smart parking. IEEE Access, 9, 61171–61183. https://doi.org/10.1109/ACCESS.2021.3074887
• Chai, T., & Draxler, R. R. (2014). Root mean square error (RMSE) or mean absolute error (MAE)? Arguments against avoiding RMSE in the literature. Geoscientific Model Development, 7(3), 1247–1250. https://doi.org/10.5194/gmd-7-1247-2014
• Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 785–794). ACM. https://doi.org/10.1145/2939672.2939785
• EarlyStopping. (n.d.). Keras. https://keras.io/api/callbacks/early_stopping/
• Er, E. E., Üneş, F., & Taşar, B. (2022). Estimating dam reservoir level change of Istanbul Alibey Dam with the fuzzy SMRGT method. Osmaniye Korkut Ata University Journal of the Institute of Science and Technology, 5(Özel Sayı), 80-95. https://doi.org/10.47495/okufbed.1033693
• Facebook. (2017). Prophet: Forecasting at scale. https://facebook.github.io/prophet/
• Fan, D., Sun, H., Yao, J., Zhang, K., Yan, X., & Sun, Z. (2021). Well production forecasting based on ARIMA-LSTM model considering manual operations. Energy, 220, Article 119708. https://doi.org/10.1016/j.energy.2020.119708
• Fu, R., Zhang, Z., & Li, L. (2016). Using LSTM and GRU neural network methods for traffic flow prediction. In Proceedings of the 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC), Wuhan, China (pp. 324–328). IEEE. https://doi.org/10.1109/YAC.2016.7804912 Graves, A., & Schmidhuber, J. (2005). Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural Networks, 18(5–6), 602–610. https://doi.org/10.1016/j.neunet.2005.06.042
• Habibi, S., & Hassanpour, S. T. (2025). An explainable machine learning framework for forecasting lake water equivalent using satellite data: A 20-year analysis of the Urmia Lake Basin. Water, 17(10), Article 1431. https://doi.org/10.3390/w17101431
• Hekimoğlu, M., Çetin, A. İ., & Kaya, B. E. (2023). Evaluation of various machine learning methods to predict Istanbul’s freshwater consumption. International Journal of Environment and Geoinformatics, 10(2), 1-11. https://doi.org/10.30897/ijegeo.1270228
• He, Z., Lin, D., Lau, T., & Wu, M. (2019). Gradient boosting machine: A survey. arXiv. https://arxiv.org/abs/1908.06951
• Huang, Z., Xu, W., & Yu, K. (2015). Bidirectional LSTM-CRF models for sequence tagging. arXiv. https://doi.org/10.48550/arXiv.1508.01991
• Huang, Z., Yang, F., Xu, F., Song, X., & Tsui, K.-L. (2019). Convolutional gated recurrent unit–recurrent neural network for state-of-charge estimation of lithium-ion batteries. IEEE Access, 7, 93139–93149. https://doi.org/10.1109/ACCESS.2019.2928037
• Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679–688. https://doi.org/10.1016/j.ijforecast.2006.03.001
• İstanbul Büyükşehir Belediyesi (İBB). (2023). Ispark parking details web service. https://ulasav.csb.gov.tr/dataset/34-ispark-otopark-detay-bilgileri-web-servisi
• Keras Team. (n.d.). Keras: Deep learning for humans. https://keras.io
• Le, X. H., Ho, H. V., Lee, G., & Jung, S. (2019). Application of long short-term memory (LSTM) neural network for flood forecasting. Water, 11(7), Article 1387. https://doi.org/10.3390/w11071387
• Li, W., Kiaghadi, A. & Dawson, C. (2021). High temporal resolution rainfall–runoff modeling using long-short-term-memory (LSTM) networks. Neural Computing and Applications, 33, 1261–1278. https://doi.org/10.1007/s00521-020-05010-6
• Nalıcı, M. E., & Akbaş, A. (2022). Forecasting of occupancy rate of dams in Istanbul. European Journal of Science and Technology, 41, 229-239. https://doi.org/10.31590/ejosat.1084484
• Pekel, E. (2023). Forecasting water demand for Istanbul by applying different machine learning algorithms. SSRN. http://dx.doi.org/10.2139/ssrn.4476915
• Qin, Y., Song, D., Chen, H., Cheng, W., Jiang, G., & Cottrell, G. W. (2017). A dual-stage attention-based recurrent neural network for time series prediction. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI-17) (pp. 2627–2633). IJCAI Organization. https://doi.org/10.24963/ijcai.2017/366
• ReduCeLrOnPlateau. (n.d.). Keras. https://keras.io/api/callbacks/reduce_lr_on_plateau/
• Retief, H., Vigneswaran, K., Ghosh, S., Andarcia, M. G., & Dickens, C. (2025). Satellite-surface-area machine-learning models for reservoir storage estimation: Regime-sensitive evaluation and operational deployment at Loskop Dam, South Africa. arXiv. https://arxiv.org/abs/2502.19989
• Sapitang, M., Ridwan, W. M., Faizal Kushiar, K., Najah Ahmed, A., & El-Shafie, A. (2020). Machine learning application in reservoir water level forecasting for sustainable hydropower generation strategy. Sustainability, 12(15), Article 6121. https://doi.org/10.3390/su12156121
• Schuster, M., & Paliwal, K. K. (1997). Bidirectional recurrent neural networks. IEEE Transactions on Signal Processing, 45(11), 2673–2681. https://doi.org/10.1109/78.650093
• Sekban, J., Nabil, M. O. M., Alsan, H. F., & Arsan, T. (2022). Istanbul dam water levels forecasting using ARIMA models. In 2022 Innovations in Intelligent Systems and Applications Conference (ASYU). IEEE. https://doi.org/10.1109/ASYU56188.2022.9925418
• Şentürk, Ö. N., Üneş, F., Demirci, M., & Taşar, B. (2024). Forecasting of dam lake water level using M5 decision tree and anfis models. International Journal of Environmental, Agriculture and Biotechnology, 9(4), 114-119. https://dx.doi.org/10.22161/ijeab.94.16
• Taylor, S. J., & Letham, B. (2018). Forecasting at scale. The American Statistician, 72(1), 37–45. https://doi.org/10.1080/00031305.2017.1380080
• TensorFlow. (2021). TensorFlow. https://tensorflow.org
• Üneş, F. (2010). Dam reservoir level modeling by neural network approach: A case study. Neural Network World, 20(4), 461-474.
• Üneş, F., Demirci, M., & Kişi, Ö. (2015). Prediction of millers ferry dam reservoir level in USA using artificial neural network. Periodica Polytechnica Civil Engineering, 59(3), 309-318. https://doi.org/10.3311/PPci.7379
• Qie, G., Zhang, Z., Getahun, E., & Mamer, E. A. (2023). Comparison of machine learning models performance on simulating reservoir outflow: A case study of two reservoirs in Illinois, U.S.A. Journal of the American Water Resources Association, 59(3), 554–570. https://doi.org/10.1111/1752-1688.13040
• Yan, H., Qin, Y., Xiang, S., Wang, Y., & Chen, H. (2020). Long-term gear life prediction based on ordered neurons LSTM neural networks. Measurement, 165, Article 108205. https://doi.org/10.1016/j.measurement.2020.108205
• Yılmaz, F., Ulusoy, I., & Toros, H. (2020). Temporal analysis of Istanbul Water Reservoir levels and suggestions for solution. Journal of Research in Atmospheric Science, 2(2), 51-55.