Analysis and Statistical Modeling of Temperature in Bushehr Station (1951-2005)

In this study, different statistical methods for realizing the behavior (trend, fluctuation, oscillation) of temperature were used in Bushehr city. Run Test with 5% confidence Interval showed that temperature annual observations are not accidental. Trend estimate using parametric and non-parametric techniques (Pearson correlation, Spearman, Man-Kendal) were done on the observations. The results showed that annual temperature has meaningful process. Spectral analysis was used for estimate the latent cycles in annual temperature. It was indicated that in reliability level of 95% except first harmonic was, harmonic 18 and 21 meaningful. Markov chain model on monthly temperature of Bushehr monthly temperature data from the two-mode Markov chain model followed and according to this model, the occurrence likelihood of warm months at Bushehr was 0.6019 and occurrence likelihood of cold months was 0.5217 and return period of warm months about 5 months and return period of cold months was about 7 month. Finally, to anticipate the annual temperature in Bushehr, ARIMA Model has been used accordingly; three models have been fitted on Bushehr temperature time series. Based on goodness of fit including the normalizing the residuals and their independence test and significant normal based on AIC criteria, The ARIMA (o, 1, 1) model was more fitable on annual temperature. Therefore, for the next 20 years, temperature in confidence interval of 95% was anticipated.