However Meteonorm does not provide any information on how they arrive at these values. Since the data outputs from Meteonorm software are combination of multiple datasets (satellite-derived and ground measured) from vicinity of selected site it is not clear how they estimate yearly variability in situation where yearly variability of one dataset may differ from the other. Also, the method used by Meteonorm for estimating uncertainty appears quite simplistic and does not give realistic results always. http://www.meteonorm.com/images/uploads/downloads/Accuracy_of_Meteonorm_7.pdf http://www.meteonorm.com/images/uploads/downloads/PVSEC_11_mn7_p.pdf For example, in some case when calculating final uncertainty, they are even considering uncertainty of ground measurements to be 1%. This is never the case (not even for research-grade BSRN stations). In reality the best achievable uncertainty for ground measurements is 2% for GHI (and in most cases uncertainty will be higher because of poor maintenance of ground sensors).
