SavedAdd your journal

Journals / Politeknik Dergisi / 2026 / Cilt: 29 Sayı: 5

Performance of LSTM-Based Models Improved with Different Optimization Methods in Daily Streamflow Forecasting

Farklı Optimizasyon Yöntemleriyle Geliştirilen LSTM Tabanlı Modellerin Günlük Akarsu Akım Tahminindeki Performansı

Hamza Arda Deveci(ERZİNCAN BİNALİ YILDIRIM ÜNİVERSİTESİ)0000-0001-9926-8768
Hüseyin Yildirim Dalkılıç(ERZİNCAN BİNALİ YILDIRIM ÜNİVERSİTESİ)0000-0002-4405-9341
Pages
1–13
Publication Date
2026-09-16
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The results showed that the standard LSTM model provided only a limited improvement over the Linear Regression model; among the optimized models, the LSTM-DE model delivered the best performance.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Reliable forecasting of streamflow is crucial for water resources planning, flood risk reduction, and supporting hydrological decision-making processes.
📊
Key Findings & Data:In this study, the performance of LSTM-based models developed using different optimization methods for daily streamflow forecasting was examined in a comparative analysis.
💡
Conclusion & Impact:The results showed that the standard LSTM model provided only a limited improvement over the Linear Regression model; among the optimized models, the LSTM-DE model delivered the best performance.
🔬Design:Modeling & Simulation
📍Scope:2011 Verileri
⚙️Variables:Günlük akım tahmini · LSTM · hiperparametre optimizasyonu

Abstract

Reliable forecasting of streamflow is crucial for water resources planning, flood risk reduction, and supporting hydrological decision-making processes. In this study, the performance of LSTM-based models developed using different optimization methods for daily streamflow forecasting was examined in a comparative analysis. The application was carried out using daily flow data from 1970 to 2011 for the Karadere flow observation station (No. 2202) in the Eastern Black Sea Basin. Linear regression was used as the reference model; in addition to the standard LSTM model, LSTM-PSO, LSTM-BO, and LSTM-DE models were created, with their hyperparameters determined using Particle Swarm Optimization, Bayesian Optimization, and Differential Evolution methods, respectively. The models were evaluated using the RMSE, MAE, R², NSE, KGE, MBE, PBIAS, and Willmott d metrics. The results showed that the standard LSTM model provided only a limited improvement over the Linear Regression model; among the optimized models, the LSTM-DE model delivered the best performance. The LSTM-DE model achieved the lowest error level with an RMSE value of 4.00661 and produced the highest results in terms of R² and NSE. However, the analysis revealed that deviation-based indicators should be evaluated in conjunction with error and fit metrics.

Özet

Akarsu akımlarının güvenilir biçimde tahmin edilmesi, su kaynaklarının planlanması, taşkın riskinin azaltılması ve hidrolojik karar süreçlerinin desteklenmesi açısından önemlidir. Bu çalışmada, günlük akarsu akım tahmini için farklı optimizasyon yöntemleriyle geliştirilen LSTM tabanlı modellerin performansı karşılaştırmalı olarak incelenmiştir. Uygulama, Doğu Karadeniz Havzası’ndaki 2202 numaralı Karadere akım gözlem istasyonuna ait 1970–2011 dönemi günlük akım verileri kullanılarak gerçekleştirilmiştir. Lineer Regresyon referans model olarak ele alınmış; standart LSTM modeline ek olarak Parçacık Sürü Optimizasyonu, Bayesyen Optimizasyon ve Diferansiyel Evrim yöntemleriyle hiperparametreleri belirlenen LSTM-PSO, LSTM-BO ve LSTM-DE modelleri oluşturulmuştur. Modeller RMSE, MAE, R², NSE, KGE, MBE, PBIAS ve Willmott d ölçütleriyle değerlendirilmiştir. Elde edilen sonuçlar, standart LSTM modelinin Lineer Regresyon modeline göre sınırlı bir iyileşme sağladığını; optimize edilen modeller arasında ise LSTM-DE modelinin en başarılı performansı verdiğini göstermiştir. LSTM-DE modeli 4,00661 RMSE değeriyle en düşük hata düzeyine ulaşmış ve R² ile NSE açısından en yüksek sonuçları üretmiştir. Bununla birlikte, sapma temelli göstergeler hata ve uyum ölçütleriyle birlikte değerlendirilmesi gerektiğini ortaya koymuştur.

🔖

Cite This Article

Deveci, H. A., Dalkılıç, H. Y. (2026). Performance of LSTM-Based Models Improved with Different Optimization Methods in Daily Streamflow Forecasting. Politeknik Dergisi, Cilt: 29 Sayı: 5, 1–13. https://doi.org/10.2339/politeknik.1981798

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (24)

  1. Rasouli K., Hsieh W.W. and Cannon A.J., “Daily streamflow forecasting by machine learning methods with weather and climate inputs”, J. Hydrol., 414-415: 284-293, (2012).
  2. Mosavi A., Ozturk P. and Chau K.W., “Flood prediction using machine learning models: Literature review”, Water, 10(11): 1536, (2018).
  3. Samui P., Yesilyurt S.N., Dalkilic H.Y., Yaseen Z.M., Roy S.S. and Kumar S., “Comparison of different optimized machine learning algorithms for daily river flow forecasting”, Earth Sci. Inform., 16(1): 533-548, (2023).
  4. Box G.E.P. and Jenkins G.M., “Time Series Analysis: Forecasting and Control”, Holden-Day, San Francisco, (1976).
  5. Maier H.R. and Dandy G.C., “Neural networks for the prediction and forecasting of water resources variables: A review of modelling issues and applications”, Environ. Model. Softw., 15(1): 101-124, (2000).
  6. Kumar D.N., Raju K.S. and Sathish T., “River flow forecasting using recurrent neural networks”, Water Resour. Manag., 18(2): 143-161, (2004).
  7. Karran D.J., Morin E. and Adamowski J., “Multi-step streamflow forecasting using data-driven non-linear methods in contrasting climate regimes”, J. Hydroinform., 16(3): 671-689, (2014).
  8. Montgomery D.C., Peck E.A. and Vining G.G., “Introduction to Linear Regression Analysis”, John Wiley & Sons, Hoboken, (2021).
  9. Shen C., “A transdisciplinary review of deep learning research and its relevance for water resources scientists”, Water Resour. Res., 54(11): 8558-8593, (2018).
  10. Torres J.F., Hadjout D., Sebaa A., Martinez-Alvarez F. and Troncoso A., “Deep learning for time series forecasting: A survey”, Big Data, 9(1): 3-21, (2021).
  11. Hochreiter S. and Schmidhuber J., “Long short-term memory”, Neural Comput., 9(8): 1735-1780, (1997).
  12. Lv N., Liang X., Chen C., Zhou Y., Li J., Wei H. and Wang H., “A long short-term memory cyclic model with mutual information for hydrology forecasting: A case study in the Xixian basin”, Adv. Water Resour., 141: 103622, (2020).
  13. Kratzert F., Klotz D., Brenner C., Schulz K. and Herrnegger M., “Rainfall-runoff modelling using Long Short-Term Memory (LSTM) networks”, Hydrol. Earth Syst. Sci., 22(11): 6005-6022, (2018).
  14. Kratzert F., Klotz D., Shalev G., Klambauer G., Hochreiter S. and Nearing G., “Benchmarking a catchment-aware Long Short-Term Memory network (LSTM) for large-scale hydrological modeling”, Hydrol. Earth Syst. Sci. Discuss.: 1-32, (2019).
  15. Xiang Z., Yan J. and Demir I., “A rainfall-runoff model with LSTM-based sequence-to-sequence learning”, Water Resour. Res., 56(1): e2019WR025326, (2020).
  16. Dalkilic H.Y., Kumar D., Samui P., Dixon B., Yesilyurt S.N. and Katipoğlu O.M., “Application of deep learning approaches to predict monthly stream flows”, Environ. Monit. Assess., 195(6): 705, (2023).
  17. Kennedy J. and Eberhart R., “Particle swarm optimization”, Proc. IEEE Int. Conf. Neural Netw., Perth, Australia, 1942-1948, (1995).
  18. Jones D.R., Schonlau M. and Welch W.J., “Efficient global optimization of expensive black-box functions”, J. Global Optim., 13(4): 455-492, (1998).
  19. Storn R. and Price K., “Differential evolution – A simple and efficient heuristic for global optimization over continuous spaces”, J. Global Optim., 11(4): 341-359, (1997).
  20. Lachure S. and Tiwari A., “Hydrological predictive modeling for Indian River: Leveraging LSTM and GRU attention mechanisms”, SN Comput. Sci., 6(6): 749, (2025).
  21. Günay A., Kumantaş M., Kayhan A. and Anılan T., “Sürü Tabanlı Algoritma Modelleri ile Taşkın Debisi Tahmini”, Politeknik Dergisi, 28(3): 825-832, (2025).
  22. Akman T., Yılmaz C. and Sönmez Y., “Short-term Electric Energy Load Forecasting of Ankara Region Using Artificial Intelligence Methods”, Politeknik Dergisi, 26(4): 1517-1531, (2023).
  23. Yapıcı A.T. and Abut N., “Geleceğe Yönelik Elektrikli Araç ve Şarj İstasyonu Sayılarının LSTM ve GRU Derin Öğrenme Yöntemleri Kullanılarak Tahmin Edilmesi: Kocaeli İli Örneği”, Politeknik Dergisi, 29(4): 1-14, (2026).
  24. Shmueli G., “To explain or to predict?”, Stat. Sci., 25(3): 289-310, (2010).