Document Type : Research Paper

Author

• Mohammad Reza Mirzaei

Assistant Professor, Faculty of Agriculture, Yasouj University, Iran

Abstract

In this study, the mean monthly air temperature data from Shah-Mokhtar hydrometric station in Kohgiloyeh and Boyer Ahmad province for a period of 39 years between 1970 and 2009 was investigated. Using different graphical EDA techniques such as spectral, autocorrelation and partial autocorrelation plots,  i) existence of seasonal part and ii) suitability of using time series analysis to model the residuals were determined. The lag plot and autocorrelation plot of the original data showed that a sinusoidal model was appropriate to model the seasonal effect. So using sinusoidal model and determining its parameters precisely, the seasonal effect was modeled properly. Time series analysis was also used to model the residuals using ARIMA models. Among different models, ARIMA(0,1,2) model was selected as the best model using Normalized Bayesian Information Criterion (NBIC). Finally, null hypotheses for Kolmogorov-Smirnov and Ljung-Box tests were not to be rejected at 5% level for the obtained model which confirms the adequacy of the model.

Keywords

• ARIMA
• Autocorrelation
• EDA
• temperature
• time series