In meteorological domain, researchers have been applying ConvNets on satellite images to accomplish different tasks. Chen et al.  proposed a ConvNet model to predict the intensity of cyclones that outperformed many contemporary meteorological methods. Larraondo et al.  used CNN models to interpret numerical weather model data. These evidently demonstrate the unlimited potential of combining ConvNets and satellite images in providing solutions to different tasks accurately, inexpensively and with high scalability. In this work, we proposed a Metadata-Augmented CNNLSTM model (in Fig.3) to provide cross-location solar irradiance nowcasting through the use of satellite images based on the observation that solar irradiation is mostly affected by cloud coverage. With the global coverage of satellite images, our proposed model can predict solar irradiance for the past, now, and near future in any locations.