An Assessment the Performance of Genetic Programming and Auto Regresive Moving Average on the Daily Discharge Prediction(Case study: the Amameh Watershed)

Shortage of water resources and the growing concern about the sustainable development have made the water supply for all of the potential needs nearly impossible.As an accurate prediction of river discharge is very important in water resources management, the development of a model to predict discharge has been carried out using the genetic programming and auto regression moving average on the Amameh Watershed located in the Province of Tehran. The long-term rainfall, temperature, discharge, relative humidity, and evaporation data have been used. Satisfactorily, the results showed that genetic programming had a lower error and could estimate the observed discharge. Furthermore, the number 54 model with inputs of temperature, rain, the delay in rainfall of up to two days, relative humidity, evaporation, and the delay in discharge of up to two days were considered as the best fit model with the errors of 0.001, 0.031 and 0.009 in the training stage and 0.002 , 0.032, and 0.009 at the testing stage respectively. On the other hand, the linear auto regression moving average models showed a much higher error; they could neither predict the high discharge, nor low flow and have not been able to provide satisfactory results. Therefore, the application of a genetic programming model is recommended due toits high precision with the main operators and the standardized data.