Markov processes and Fourier analysis as a tool to describe and simulate daily solar irradiance

Abstract Time series of 20 years of daily solar irradiance data from four italian stations are analyzed on a statistical basis. It is shown that the irradiation sequences are not stationary, both in the mean and in the variance. They can be determined by three components: (a) a mean, well described by a Fourier series with only one coefficient; (b) a variance about the mean, well fitted by a Fourier expansion with two coefficients; (c) a stochastic component. The stochastic component follows a first order Markov model. Since it has a non-normal distribution, a normalizing transform has been introduced which does not affect its statistical properties.