The temporal analysis of solar radiation recorded on the ground provides information on atmospheric opacity concerning the propagation of solar radiation. This is particularly useful for various applications such as photovoltaic installation systems, ground-space optical communications, and even studies related to the Earth’s atmosphere. This paper focuses on the estimation of global solar irradiance at ground level from satellite observations using the statistical method Heliosat-2. This work is carried out using one year (2015) of MSG satellite images from its 0.6, 0.8, and HRV measurement channels. The model ESRA2 is retained for the study. The ground data for 2015 used for the development of the method were collected at the site of Ghardaïa located in the south of Algeria, which has a radiometric station. The obtained results show that ESRA2 associated with the statistical model is well-suitable to estimate global solar radiation (GSR) on the ground from the images of the HRV channel compared to those of the other channels. Indeed, an RMSE of 612.57 Wh/m2, a MAPE of 9.24%, an MBE of 128.69 Wh/m2, and a correlation coefficient R of 93% were obtained for the HRV channel. These statistical errors and the correlation coefficient were found respectively equal to 711.15 Wh/m2, 10.80, 156%.96 Wh/m2, and 91% for the near-infrared channel 0.8 µm of MSG equal to 671.34 Wh/m2, 10.31%, 193.72 Wh/m2, and 92% for the visible channel 0.6 µm.