Fitting the Seasonal Time Series Model to the rivers discharge in time domain (Case Study: Atrak River)

Nowadays, investigating forecasting of the hydrological variables behavior and the effective climatological factors on it in the time domain are considered by researchers. Therefore, the usage of time dependent data analysis in the prediction of river discharge rates is statistically valid, and use them for water resources management and infrastructure designing is an inevitable issue. Atrak River is one of the most important water resources in economic, agricultural and environmental issues in the North East of Iran. In recent years, this river had overflow or decline the water levels that certainly affected on river discharge rates. In this paper, we use the data of the six hydrometric stations from 1352-1382 of Atrak river. In this paper, first we have a descriptive analysis of data and then fit suitable models for discharge time series data. At last, with estimate the parameters of the model, predict the distribution of the river flow rate. In fit model and estimate the parameters of the model, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and Mean Square Error (MSE) is important. After fix the variance of the data and remove the trend and seasonality variation and use the Autocorrelation and Partial Autocorrelation plots, we choose the ARIMA(1,0,2)(1,1,2)12 model. At last, with this time series model, we forecast the rate of the river discharge for next year and observe that it will decline in the coming year.