Satellite-derived solar radiation products often have strong deviations from ground-measured solar irradiance data. Common problems include a systematic over- or underestimation, which can be expressed by calculating the mean bias. As in first order energy produced by a solar plant roughly depends on the total incoming irradiation this simple number gives a first hint on potential error of energy production. Moreover, due to limited usability of low and very high irradiance values and non-linear effects also the frequency distribution of solar time-series data have a relatively strong influence on potential yields and its uncertainty. Thus, for detailed engineering and financing of solar power plants it is advisable to not rely simply on site-specific satellite derived solar resources products, but also to do measurements at the site and use those for validating and improving the satellite-derived time-series. This paper presents a method, which embosses the features of frequency distribution of ground-based measurements onto two different satellite-derived time-series. The method presented shows good improvement of systematic over- or underestimation: in one example it is shown, that if at least 3 months of parallel measurements are available at a site, even strong biases beyond 25 % can be reduced below 10 %. But the quality of results applying only a quarter year of overlap for training is still quite variable. Applying half a year of data usually already gives significantly better results - at least if all sun angles are covered, by a start of measurements at summer or winter solstice. In most cases, a complete parallel year of measurements is sufficient to reach very good improvements at sites. Inter-comparing the improvement effect on beam irradiance versus global horizontal irradiance shows that direct normal values from satellite can be far better improved than on global horizontal irradiance. However, the initial analysis shows that the need to correct global horizontal is also much smaller.
|Title of host publication||30th ISES Biennial Solar World Congress 2011, SWC 2011|
|Number of pages||12|
|Publication status||Published - 2011|
|Name||30th ISES Biennial Solar World Congress 2011, SWC 2011|
- Mean bias
- Solar irradiance