Solar Position Identification on Sky Images for Photovoltaic Nowcasting applications

Nowcasting, i.e. very-short term forecasting, of photovoltaic (PV) production has a growing importance. A typical method to perform it, relies on sky cameras and sky images that, duly processed, can provide a forecast of sharp and fast changes in PV production. The first issue that arises while dealing with these analyses regards the identification and forecasting of the solar position in the images, especially when it is covered by clouds. In this paper, three different techniques for solar identification in sky images are proposed, analysed and compared.