Short-term and long-term streamflow forecasting using a wavelet and neuro-fuzzy conjunction model

Streamflow forecasting is an important issue in hydrologic engineering, as it determines the reservoir inflow as well as the flooding events, in spite of several other applications in water resources engineering. In the present study, the application of hybrid wavelet-neuro-fuzzy model has been investigated to model daily, monthly and yearly streamflows. Streamflow data of Derecikviran Station on the Filyos River in the Western Black Sea region of Turkey were used in the study. The data sample consisted of 31 years of streamflow records. In the first part of the study, single neuro-fuzzy (NF) and wavelet-neuro-fuzzy (WNF) models were established based on the previously recorded streamflow values and compared with each other. It was found that the WNF model increase the accuracy of the single NF models especially in forecasting yearly streamflows. In the second part of the study, the single NF and WNF models were compared with each other by adding periodicity component into the their inputs. The comparison results indicated that adding periodicity component generally increased the models’ accuracy.