In this study the wavelet-neuro-fuzzy model, which combines the wavelet transform and the neuro-fuzzy technique, has been employed to forecast monthly streamflows. The observed monthly streamflow data are decomposed into some sub-series (components) by discrete wavelet transform and then appropriate sub-series are used as inputs to the neuro-fuzzy models for forecasting monthly streamflows. The data from two stations, Durucasu and Tanir, in Turkey are used as case studies. The wavelet-neuro-fuzzy forecasts are compared with those of the single neuro-fuzzy models. Comparison results indicate that the wavelet-neuro-fuzzy model is superior to the classical neuro-fuzzy method especially for the peak values. For the Durucasu and Tanir stations, it was found that the wavelet-neuro-fuzzy models are superior in forecasting monthly streamflows than the optimal neuro-fuzzy models.
Research Article|December 01 2011
Wavelet and neuro-fuzzy conjunction model for streamflow forecasting
Özgür Kişi, Turgay Partal; Wavelet and neuro-fuzzy conjunction model for streamflow forecasting. Hydrology Research 1 December 2011; 42 (6): 447–456. doi: https://doi.org/10.2166/nh.2011.048
Download citation file: