The Application of Seasonal Autoregressive Fractionally Integrated Moving Average (SARFIMA) in Forecasting of River Streamflow

Abstract

Time series modeling can be used in various fields including hydrology. River streamflow is one of
the hydrological parameters which is not only affected by seasonal factors but also often identified to possess
long memory pattern. In this paper, a modeling using Seasonal Autoregressive Fractionally Integrated Moving
Average (SARFIMA) will be applied. The data used is historical data of Cimanuk river streamflow which is
the result of 20-year documentation in monthly interval. SARFIMA model is then compared with ARFIMA.
The analysis is done to comprehend how SARFIMA model is able to model seasonal factors and long memory
pattern which is shown by the data of Cimanuk river streamflow. The result of the analysis shows that
SARFIMA model is not suitable for this data based on MSE and MAPE value.
Key words: Sarfima, seasonal factors, MAPE.