The data set comprises half-hourly values of global, direct normal, and diffuse irradiance with a spatial resolution of 1 km x 1 km at the sub-satellite point. The spatial resolution varies somewhat with the distance to the sub-satellite point depending on the viewing angle of the satellite. However, independent of viewing geometry, the resolution is always better than 2 km.

Satellite derived estimates of solar energy have been extensively tested against ground measurements with pyranometers. It was found that satellite data unfold their strength in mid to long-term estimates where the time period ranges from a day to a couple of years. The longer the time period over which energy is accumulated, the more satellite images are available, and the better is the accuracy. For daily values of global irradiance the accuracy is normally in the range of 8-10% comparable to the accuracy of solar cell radiation sensors. A monthly irradiance estimate has an error of only 5% and annual irradiance values reach an accuracy of 2-3% coming close to the precision of pyranometers.

The quoted error margins correspond to normal circumstances, but there are few exceptions where satellite estimates can fail almost entirely. These special cases occur for highly reflecting ground surfaces such as snow and salt-beds. The high reflectivity is falsely interpreted as being due to clouds resulting in untrustworthy irradiance estimates. These topics are subject of ongoing research.

However, as long as our users do not live in snow-covered areas or in the vicinity to a salt-bed in the desert, there are no problems. In the absence of snow, the accuracy is as good as stated above. Snow-covered regions show up as false minima in solar irradiance maps, typically confined to high-altitude areas in mountain ranges. A snow detection routine is part of our algorithm, yet, we cannot guarantee that it treats the effects on irradiation absolutely perfectly.

The snow detection routine is applied for the United States, but not in Canada. This is because we receive data on snow coverage from the US weather service which do not cover territory outside the US border. Therefore irradiance estimates for the Canadian provinces in the winter period shall be considered with some degree of limitation.