Tools for PV (photovoltaic) plant operators: Nowcasting of passing clouds

• Nowcasting of passing clouds is modeled by using a 15 s lag database.
• ARIMA (Autoregressive Integrated Moving Average) (0,1,0) model is mostly recommended for nowcasting of passing clouds.
• Models accuracy increases by increasing the radiative regime stability.

The response time of a PV (photovoltaic) plant is very short and its output power follows the abrupt change in solar irradiance level due to alternate shadow by clouds. The sunshine number (SSN) is a Boolean quantity stating whether the sun is covered by clouds or not, thus being an appropriate parameter to predict the occurrence of direct solar radiation at ground level. Various ARIMA (Autoregressive Integrated Moving Average) models for SSN nowcasting are inferred and discussed in this paper. Actinometric and meteorological data measured at 15 s lag during June 2010 in Timisoara (Romania) are used. The forecasting accuracy is studied as a function of season, of the procedure used to obtain a binary time series and of the type of white noise distribution, respectively. It is demonstrated that the ARIMA(0,1,0) model forecasts SSN with the same accuracy as higher order ARIMA models. The forecasting accuracy decreases when the instability of the radiative regime increases.