Use of autoregressive models to generate a series of daily averaged point cloudiness values

Analysis of observed data for 10 years in two Romanian localities showed that the daily averaged point cloudiness in a given day mainly depends on the cloud cover amount from the past two days. Hence, first or second order autoregressive (AR) processes can be used to generate a point cloudiness time series. The second order models are slightly better than the first order models. The AR models generate data whose mean and standard deviation are close to those of the observed data. Good agreement between the skewness of the observed and synthetic data occurs during the warm season. No concordance was emphasized between the kurtosis of observed and generated daily averaged point cloudiness values.