Analysis and prediction of water level fluctuations in Urmia Lake using ARIMA model

This study has been done to evaluate the fluctuations of water level in Urmia Lake and to provide a best model for prediction the water level fluctuations. Monthly water level data for the period (1345 - 1392) was used and homogeneity was assessed by Run Test. Then the stability of mean and variance of the data was tested in order to put down the non-stability by creation a rank in series. Trend of the monthly series was eliminated by making a difference and the time series of water level was evaluated by using Box- Jenkins model and the best model was fitted. Accuracy of the model was verified based on AIC, BIC and chart analysis of autocorrelation and partial autocorrelation functions and ARIMA = (0, 1, 4) (1, 1, 1)12 was selected as a suitable model. The selected model was fitted then the model was tested by Analysis of residuals and confirmed its authenticity. Finally, the monthly behavior of the series was predicted for 9 years later by using this model.