Abstract
Stochastic–dynamic models are discussed for both air temperature and solar irradiance daily time series in the Italian climate. Most of the methodologies discussed in this paper are well known and established for processes having a Gaussian distribution. However, a technique is presented that allows statistical inferences for non-Gaussian processes. Applying these models to 20-year time series, their predictability is analyzed for five meteorological stations of the Areonautica Militare Italiana. The following results were obtained: 1) the seasonalities in both the mean and the standard deviations of measured data are well fitted by simple periodic models; 2) the short range statistical fluctuations of the analyzed variables are well described by first order autoregressive processes whose parameters have constant values for all five stations. For the sake of brevity results are presented only for one station (Napoli).
