An Evaluation of Parameters Estimation Methods for ARMA Model in order to Modeling and Predicting Annual Rainfall: A Case Study of Shahrekord Synoptic Station

Modeling of hydrological processes is complicated and important due to high number of parameters and interaction between them. So it is important to choose suitable model and appropriate estimation methods of such models’ parameters. In this study, annual rainfall for 10 years modeled and forecasted using measured data at Shahrekord synoptic station during 1961-2010 and linear ARMA time series models. To do this, two methods of moments and maximum likelihood applied to estimating ARMA model parameters. Both methods confirm ARMA (0,1) model based on the least Akaike criterion. Results related to evaluation and verification of model showed that moments method predicted the time series more accuate than maximum likelihood method.