Solar radiation data are necessary to estimate the photovoltaic power generation. To estimate solar radiation at any given location, meteorological satellite images have been used to this time. However, accurate estimations under specific weather conditions, such as slightly cloudy skies, were difficult due to computational costs and the complex effects of cloud structure on radiative transfer. In this study, a solar radiation estimation model was developed using multiband information of Himawari-8 by combining a neural network and a radiative transfer model. The validation results showed that the newly developed model improved overall estimation accuracy and was able to adequately represent the solar radiation under certain conditions.
