The Sun is the primary source of energy for the Earth. There is a strong need for information on solar radiation in many domains, but there is a large discrepancy between user request and available information. Solar radiation is measured by ground networks, but measurements are punctual and scattered. Therefore interpolation or extrapolation techniques are necessary, but they are leading to errors growing with the distance from the station. In this framework, we are proposing firstly a method to estimate solar radiation from satellite images, and secondly tools to allow the customer to access more efficiently to information on solar radiation throughout Internet. Among the existing methods, the Heliosat method has been chosen with regard to quality, robustness, and exploitation criteria that must be stronger considering the objective of creating a climatology of solar radiation. We are thus proposing sorne changes to improve the Heliosat method by 1) introducing a method for the automatic calibration of Meteosat sensors, 2) using a more accurate clear-sky model, 3) establishing a new relationship between cloud and clear-sky indices, 4) evaluating albedos of various surfaces from calibrated images. Consequently, the "Heliosat-2" method will be a strong basis to elaborate a climatic database of solar radiation to be disseminated throughout the Internet.
