Predicting monthly streamflow using a hybrid wavelet neural network: Case study of the Çoruh river basin

Creative Commons License

Güneş M. Ş., Parim C., Yıldız D., Büyüklü A. H.

Polish Journal of Environmental Studies, cilt.30, sa.4, ss.3065-3075, 2021 (SCI-Expanded) identifier identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 30 Sayı: 4
  • Basım Tarihi: 2021
  • Doi Numarası: 10.15244/pjoes/130767
  • Dergi Adı: Polish Journal of Environmental Studies
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, BIOSIS, CAB Abstracts, Central & Eastern European Academic Source (CEEAS), Environment Index, Greenfile, Public Affairs Index, Veterinary Science Database
  • Sayfa Sayıları: ss.3065-3075
  • Anahtar Kelimeler: streamflow, artificial neural network (ANN), wavelet transform (WT), air temperature, precipitation, CLIMATE-CHANGE IMPACTS, MOVING AVERAGE, LAND-USE, RAINFALL, RUNOFF, PRECIPITATION, TEMPERATURE, TRANSFORM, ENSEMBLE, FLOW
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

© 2021, HARD Publishing Company. All rights reserved.In this study, a hybrid model combining discrete wavelet transforms (WTs) and artificial neural networks (ANNs) is used to estimate the monthly streamflow. The WT-ANN hybrid model was developed using the Daubechies main wavelet to predict the streamflow for three gauging stations on the Çoruh river basin one month in advance, with different combinations of air temperature, precipitation, and streamflow variables, and their wavelet transformations. Four different hybrid WT-ANN models were generated and compared with four different conventional ANN models. The dataset was chronologically divided into training, validation, and testing data. The results indicated that the WT-ANN hybrid models performed better than the traditional ANN models for all three stations. Furthermore, the chronologically divided dataset was used to examine the effects of changes in hydrological data over time on model performance. In conclusion, model performances in the training period deteriorated during the validation and testing periods due to structural changes in the hydrological data.