Solar irradiance is the "fuel" of solar power plants. The efficient use of solar radiation for solar energy applications requires a detailed knowledge of the short- and long-term variation of the radiation supply. Hourly values are necessary to optimize solar energy installations. Reliable estimates of solar energy yields can only be achieved by analysing multi-year series in high resolution. At IPA we use our knowledge in Earth observation and radiative transfer modeling to address these issues. We have developed two databases of direct and global solar irradiance for solar energy applications: DLR-ISIS, a 21-year data set with low spatial resolution which is freely available for scientific purposes; and the commercial high resolution SOLEMI data set. For the operation of concentrating solar power plants we work on improving the prediction of solar radiation with numerical weather prediction models. New developments of satellite and model-based methods for mapping and forecasting of solar energy yields are done in the framework of the Helmholtz virtual Institute of Energy Meteorology (vIEM) in collaboration with University of Oldenburg, DLR-DFD and DLR-ITT.