Stage -Discharge Relationship Modeling in Rivers Using Intelligent Systems

Since continual measurement of rivers discharge, specialy at flooding time is expensive and difficult task, modeling of stage-discharge relationship using mathematical and intelligent models have been developed and used. In this research artificial neural networks, inference neuro-fuzzy and genetic programming were used for modeling daily stage- discharge relationship at the two stations, Yamula and Sogutluhan on the Kizilirmak river in Turkey. Modeling were accomplished by the above three methods with various combination of inputs including the previous stage-discharge data. According to the results obtained, neuro-fuzzy showed greater accuracy. Due to ability of genetic programming for evolving mathematical relationship and selecting significant variables, this method was selected as the best for stage-discharge relationship modeling using four main operators including {+,-,*,/} with R2,RMSE and MAE for Yamula station are 1.000, 0.974 and 0.552 and for Sogutluhan station, 0.999, 1.095 and 0.630 respectively.