Monthly Precipitation Forecast by Time Series Packages in R Environment (Case study: Arazkooseh station of Golestan province)

Water shortage or its improper distribution is one of the biggest concerns in this century، and the precipitation is one of the most important water resources. Therefore، predicting of the precipitation at each catchment is one of the crucial climate quantities. To forecast precipitation، time series analysis can be used. One of the statistical software that can perform these calculations easily and with high accuracy is R software which was used in this study. In this research monthly precipitation for 30 years was considered as a time series in Arazkooseh station، Golestan province. Based on the autocorrelation and partial autocorrelation diagrams، the appropriate model ARIMA (0،0،1) (0،1،1) 12 was fitted to the considered data set. Further the residual autocorrelation diagrams for the fitted model show that the residuals are uncorrelated. Except at the lag 11th where Bartlett test showed correlation is zero (α=0. 05)، there were correlation at the other lags. Furthermore the P-value of the Ljung-Box test statistic for different lags is greater than the significance level of 5%، which indicates that the residuals are uncorrelated. Moreover the precipitation for the next 12 months was forecasted based on the fitted model.