The use of time series modeling for the determination of rainfall climates of Iran

In this study, regional climates of Iran were identified based on the properties of the monthly rainfall time series models of 28 main cities of Iran. The autocorrelation (ACF) and partial autocorrelation (PACF) of selected series revealed the seasonal behavior of the monthly rainfall. After the parameters of the models were estimated and the residuals of the models analysed to be time independent and the normality was checked using Portmanteau lack of fit and nonparametric tests, the multiplicative ARIMA model was fitted to monthly rainfall time series of the stations. To determine regional climates, a hierarchical cluster analysis was applied on autocorrelation coefficients at different lags and three main climatic groups were found based on the time series models, namely, simple, moderate and complex climates. The results of the time series modeling showed a high variation of the temporal pattern of the monthly rainfall over Iran except for the margins of the Caspian Sea and the Persian Gulf. The study also shows that the correlation between the seasonal autocorrelation coefficient of the rainfall time series and the rainfall coefficient of variation and elevation of the stations is significant while lag-one autocorrelation coefficient does not correlate to rainfall coefficient of variation and the elevation of the stations. Different models also imply the high variation in the spatial rainfall producing mechanism and different stationarity and periodicity characteristics of the rainfall temporal pattern over Iran. A nomenclature of the abbreviation is given at the end of the paper. Copyright (c) 2006 Royal Meteorological Society