Araştırma Makalesi

Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models

Sayı: 6 20 Ekim 2026
PDF İndir
TR EN

Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models

Öz

Accurate forecasting of streamflows is crucial for water resources management, flooding, and climate-resilient hydrology. However, the nonlinear and nonstationary properties of hydrological series often limit the forecasting capabilities of conventional data-based techniques. This study presents a hybrid wavelet-machine learning framework for the prediction of monthly streamflows in the Karamandere Creek Basin (Türkiye) based on the combination of Discrete Wavelet Transform (DWT) and five supervised machine learning techniques, including Decision Tree (DT), Linear Regression (LR), Random Forest (RF), Extra Trees (ET), and Support Vector Regression (SVR). The effect of feature representation through multiple resolution levels on the quality of predictions was investigated for two wavelet families, Daubechies (db4) and Symlet (sym8). Hydro-meteorological daily measurements like streamflow, precipitation, and temperature obtained from 2005 to 2020 were aggregated to monthly values and further sequentially partitioned into training (154 samples) and test (36 samples) datasets in order to preserve time consistency. Model efficiency assessment was performed using the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). Experimental results indicate that feature extraction based on wavelet transforms increases the prediction precision for all machine learning algorithms in comparison with raw data inputs. In comparison with the considered approaches, the Random Forest approach coupled with Symlet wavelet features demonstrated the best predictive performance (R² = 0.92), slightly outperformed by the Extra Trees approach (R² = 0.91). Among other techniques, the Support Vector Regression technique displayed the greatest improvement rate in prediction accuracy due to the wavelet transformation of features, as the R² score increased from 0.65 to 0.88. The comparison proves the superiority of the Symlet wavelets in terms of signal decomposition compared to Daubechies wavelets because of their smoothness and interpretability. Thus, the proposed approach confirms that the multiresolution approach to feature extraction helps enhance the robustness and generalization capacity of machine learning models for predicting the streamflow.

Anahtar Kelimeler

Kaynakça

  1. [1] Doğan, S., Özçelik, S., Dolu, Ö., & Erman, O. (2010). Küresel ısınma ve biyolojik çeşitlilik. İklim Değişikliği ve Çevre, 3, 63–88.
  2. [2] Türkeş, M. (2000). Hava, İklim, Şiddetli Hava Olayları ve Küresel Isınma. T.C. Başbakanlık Devlet Meteoroloji İşleri Genel Müdürlüğü, 187, 205.
  3. [3] The Global Goals. (2025). Clean water and sanitation. https://globalgoals.org/goals/6-clean-water-and-sanitation.
  4. [4] Birleşmiş Milletler Türkiye. (2025). Sürdürülebilir Kalkınma Amaçları 13: İklim eylemi. https://turkiye.un.org/tr/sdgs/13.
  5. [5] Lankford, B., Bakker, K., Zeitoun, M., & Conway, D. (Eds.). (2013). Water security: Principles, perspectives and practices. Earthscan/Routledge.
  6. [6] UN-Water. (2013). What is water security? [Infographic]. https://www.unwater.org/waterfacts/water-security.
  7. [7] Jepson, W., Budds, J., Eichelberger, L., Harris, L., Norman, E., O'Reilly, K., & Young, S. (2017). Advancing human capabilities for water security: A relational approach. Water Security, 1, 46–52. https://doi.org/10.1016/j.wasec.2017.07.001.
  8. [8] Meissner, R. (2016). Water security in Southern Africa: Discourses securitising water and the implications for water governance and politics. In C. Pahl-Wostl, A. Bhaduri, & J. Gupta (Eds.), Handbook on Water Security (pp. 280–300). Edward Elgar. https://doi.org/10.4337/9781782548010.00025.

Ayrıntılar

Birincil Dil

İngilizce

Konular

Su Kaynakları Mühendisliği

Bölüm

Araştırma Makalesi

Yayımlanma Tarihi

20 Ekim 2026

Gönderilme Tarihi

18 Ağustos 2026

Kabul Tarihi

9 Ekim 2026

Yayımlandığı Sayı

Yıl 2026 Sayı: 6

Kaynak Göster

APA
Akbulut, U., Farhang, Y., Çifçi, M. A., Özyavaş, A., Aslan, Z., & Koçak, K. (2026). Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models. Turkish Journal of Hydraulic, 6. https://doi.org/10.68369/TJH2026102
AMA
1.Akbulut U, Farhang Y, Çifçi MA, Özyavaş A, Aslan Z, Koçak K. Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models. THD / TJH. 2026;(6). doi:10.68369/TJH2026102
Chicago
Akbulut, Uğur, Yousef Farhang, Mehmet Akif Çifçi, Adem Özyavaş, Zafer Aslan, ve Kasım Koçak. 2026. “Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models”. Turkish Journal of Hydraulic, sy 6. https://doi.org/10.68369/TJH2026102.
EndNote
Akbulut U, Farhang Y, Çifçi MA, Özyavaş A, Aslan Z, Koçak K (01 Ekim 2026) Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models. Turkish Journal of Hydraulic 6
IEEE
[1]U. Akbulut, Y. Farhang, M. A. Çifçi, A. Özyavaş, Z. Aslan, ve K. Koçak, “Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models”, THD / TJH, sy 6, Eki. 2026, doi: 10.68369/TJH2026102.
ISNAD
Akbulut, Uğur - Farhang, Yousef - Çifçi, Mehmet Akif - Özyavaş, Adem - Aslan, Zafer - Koçak, Kasım. “Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models”. Turkish Journal of Hydraulic. 6 (01 Ekim 2026). https://doi.org/10.68369/TJH2026102.
JAMA
1.Akbulut U, Farhang Y, Çifçi MA, Özyavaş A, Aslan Z, Koçak K. Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models. THD / TJH. 2026. doi:10.68369/TJH2026102.
MLA
Akbulut, Uğur, vd. “Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models”. Turkish Journal of Hydraulic, sy 6, Ekim 2026, doi:10.68369/TJH2026102.
Vancouver
1.Uğur Akbulut, Yousef Farhang, Mehmet Akif Çifçi, Adem Özyavaş, Zafer Aslan, Kasım Koçak. Wavelet-Enhanced Machine Learning Framework for Monthly Streamflow Prediction: A Comparative Assessment of Hybrid Models. THD / TJH. 01 Ekim 2026;(6). doi:10.68369/TJH2026102

"Türk Hidrolik Dergisi"nin Tarandığı INDEX'ler 

(Indexes : Turkish Journal of Hydraulic)
download?token=eyJhdXRoX3JvbGVzIjpbXSwiZW5kcG9pbnQiOiJqb3VybmFsIiwib3JpZ2luYWxuYW1lIjoiaW1hZ2UucG5nIiwicGF0aCI6ImY0ZjcvZTJkMS9hOGZiLzZhYjEwOTdiNDg2OTY4Ljg5NDI4NTkxLnBuZyIsImV4cCI6MTc4OTk5MDc5NSwibm9uY2UiOiIxMWIxYzFhN2EyMDZmOTMxMGE1OTU0ZTE0MjQ2MzgyZSJ9.YF0Sarxk_C3axPaRniBCTj-7gHerPnR96Ipx-3fTgMw        18820       18821      

 18985             18822      DRJI Indexed Journal            18823                18824