عنوان مقاله [English]
For over three decades, hydrologists were recommended multivariate models to describe and modeling complex hydrological processes. While recently the multivariate models in hydrology is discussed. In multivariate models, the modeling and predicting various parameters can be improved by involving other factors. In this study, univariate and contemporaneous multivariate ARMA models (CARMA) were evaluated for modeling of Urmia lake water level. The time series of Urmia Lake water level in annual scale in the period of 1982-2011 were used for ARMA models and the time series of Shahrchai, Nazloochai and Barandoozchai flow rates and Urmia Lake water level in mentioned data period were used for CARMA models. The results of evaluation and verification of models showed that by adding river flow data, the accuracy of modeling and verification of models will increase. Also the results showed that according to R-square coefficient equal to 0.75 between validation data of models and the root mean square error criterion equal to 0.62, the CARMA models can provide better results than ARMA models. Using multivariate models in the modeling and forecasting of Urmia lake water level increased the accuracy of modeling about 20 percent.