Expert quality control of solar radiation ground data sets

Ground-based radiation measurements are required for all large solar projects and for evaluating the accuracy of solar radiation models and datasets. Ground data almost always contain low-quality periods caused by instrumental issues, logging errors, or maintenance deficiencies. Therefore, quality control (QC) is needed to detect and eventually flag or exclude such suspicious or erroneous data before any subsequent analysis. The few existing automatic QC methods are not perfect, thus expert visual inspection of the data is still required. In this work, we present a harmonized QC procedure, which is a combination of various available methods, including some that include an expert visual inspection. In the framework of IEA PVPS Task 16, these tests are applied to 161 world stations that are equipped with various radiometer models, and are candidates for an ongoing benchmark of irradiance datasets derived from satellite or weather models. Because the implementation of these methods by experts, and their subsequent decisions, might lead to different QC results, the independently obtained results from nine evaluators are compared for two test sites. The QC results are found similar and more stringent than purely automated tests, even though some deviations exist due to differences in manual flagging.