Validation of models that estimate the clear sky global and beam solar irradiance

•Clear sky models are key points for the derivation of solar radiation data.•Seven clear sky models are compared with more than 30’000 site-year hourly data.•Atmospheric aerosol and water vapor content are critical inputs to clear sky models.•First order statistics are used to evaluate the clear sky model estimates. The optimal utilization of solar energy requires a thorough characterization of the solar resource. The most accurate way is to measure that resource in situ. However accurate measurements are not a common commodity, especially over longer time spans. To circumvent the lack of ground based measurements, models can be applied to estimate solar irradiance components. A fundamental component is clear sky irradiance. In particular, clear sky irradiance is used as the normalization function in models that convert meteorological satellite images into irradiance, or in models that decompose global irradiance into diffuse/direct fraction. It is therefore important to evaluate and validate clear sky irradiance models. This paper presents the results of a validation of hourly clear sky models spanning up to eight years. The validation relies on high quality measurements at 22 locations in Europe and around the Mediterranean region. Seven models are evaluated. They were selected on the basis of their published performance, their simplicity of use, and/or their computational speed; two different sources of the aerosol load are used as input to the models. The three best models show a low bias and a standard deviation ranging from ±3% to ±5%. The standard deviation of the bias across the 22 locations is of the same order of magnitude. The observed bias patterns can be largely traced to inaccuracies inherent to the sources aerosol optical depth. No particular seasonal effects are noted. A consistent limitation across all selected models, even if their direct irradiance performance can be judged satisfactory based on the standard deviation metric, is that they tend to fall short of observations for a given clear sky global clearness index value.