Streamflow time series modeling of Zayandehrud river

Abstract

Multiplicative seasonal autoregressive integrated moving average models are appropriate for the monthly stream flow of the Zayandehrud River in western Isfahan province, Iran, through the Box and Jenkins time series modeling approach. Among the selected models interpreted from ACF and PACF, seasonal multiplicative ARIMA (1,1,0) (0,1,1) satisfied all tests and showed the best performance. Seasonal moving average parameter in the model indicates periodicity, and long memory in the streamflow, while a nonseasonal autoregressive parameter indicates the linearity of the monthly streamflow. The model forecasted streamflow for 24 leading months showed the ability of the model to predict and forecast statistical properties of the streamflow.