Wind Speed Forecasting

This paper aims at providing a general class of stochastic models for hourly averagewind speed time series taking into account all the main features of wind speeddata, namely autocorrelation, non-Gaussian distribution, seasonal and diurnal nonstationarity.It will be shown that the methodology developed in this study, testedusing the data recorded in two sites of Italy, attains valuable results in terms both ofmodelling and forecasting.