• Compilation of 14 ground-measured datasets with 2147 stations worldwide. • Proposal of a guideline for standardized quality control procedures. • Evaluation of availability, quality, and characteristics of solar radiation data. • Assessment of the significance of in-situ data for global solar potential analyses. • Application of time–frequency analysis to identify features in solar radiation. Accurate solar radiation data are fundamental for solar energy research. Reliable databases are essential to quantify their spatiotemporal variability. Although long-term ground-based radiometric measurements are considered the most reliable source of global horizontal irradiance data, they are prone to inaccuracies and often lacking in many regions. This study analyzes hourly global horizontal irradiance data from 14 datasets spanning more than 50 countries to assess data availability, data quality, and statistical as well as spatiotemporal characteristics. A comprehensive stepwise quality control procedure reviews existing and new quality tests to identify suspicious data points or entire implausible time series. After quality control, 1,618 out of 2,147 stations were retained: 289 with 2–5 years, 806 with 6–15 years, and 523 with more than 15 years of reliable data in the study period 1991–2020. This highlights that measurement sites with persistent long-term, high-quality data are scarce. While analyzing long-term changes and trends remains challenging, time–frequency analysis enables estimation of the time scales contributing most to variability at each location. Depending on the site, 47.2 % to 75.2 % of total variance is explained by deterministic cycles at the semi-daily, daily, and annual scales. The contribution of different time scales primarily depends on latitude, but is also influenced by topography and local conditions. Insights gained from quality-controlled ground-based data can be combined with global reanalysis and satellite data to better characterize spatiotemporal variability in global solar radiation and improve solar energy potential estimations from local to global scale.
Sander et al. (Wed,) studied this question.