Forecasting variation of solar radiation and movement of cloud by sky image data

This paper describes a forecasting method for photovoltaic (PV) power by using the sky image data. The presented method is able to forecast the PV power one or two hours earlier at the PV generation site. The authors have forecasted and estimated the short period variation of solar radiation by analyzing sky image data as observed by a camera. To establish the forecasting methodology for a short-period variation of the PV power, the variation in solar radiation and cloud movement were estimated using sky image data. In this study, we used the sky image obtained from a camera with a fish-eye lens, a ceilometer (quantity and height of cloud), and a pyranometer installed on the roof of the building in our campus and neighboring areas.