عنوان مقاله [English]
Like many other hydro-climatological data, runoff has seasonal variability incorporated with random processes. Previous research has shown that stochastic models are the most suitable simulation tool for random variables with seasonal variability. In this study, a time series analysis approach was utilized to obtain monthly, bimonthly, and seasonal runoff stochastic models in a few sub-basins in Dez and KarunBasin, in southwestern Iran. These sub-basins vary widely in area (from 37 to 9900 square kilometers) in order to study the models with respect to the different spatial scale. The results have shown that the kind of stochastic model in longer temporal scales is not correlated with the area of the basin. Therefore, this result could be generalized to other similar basins. On the other hand, the Monthly Autoregressive Integrated Moving Average model has different patterns regarding the spatial scale of the basin. It is shown that the autoregressive order of small basins could be greater than one. Extracted stochastic models in this research can be used for runoff forecasting in future studies and research.