Geographical Research

Because of the increasing importance of supplying water for the country، water resource management is of paramount importance. Predicting precipitation، as one of the most important climatic parameters، is especially important in using water supplies. Time series can be used to predict precipitation. Time series analysis seems to be a suitable tool for such forecasting. The present work studies the comparison of the rainfall forecast precision by Box-Jenkins Time Series at three stations in Khuzestan province. The monthly rainfall data of three Meteorological stations in province for 48 years، i.e. (1961-2008) were used. The accuracy and precision of models are evaluated based on AIC statistics and the analysis of the graphs of Autocorrelation Function and Partial Autocorrelation Function. The obtained suitable seasonal rainfall model was (0،1،2)*(2،1،0)12 for Ahwaz، (0،1،2)*(0،1،1)12 for Abadan and (1،1،2)*(0،1،1)12 for Dezful. The results of this study show that these models accurately predict seasonal rainfall almost well.