LAUSR

Abstract Accurate modeling of solar radiation under cloudless situations is important to facilitate the development and exploitation of the renewable solar resource worldwide. However, many existing clear-sky solar radiation models perform inconsistently under varying aerosol conditions. This work presents a worldwide review of some of the most widely known clear-sky solar radiation models and shows that most of these inconsistencies are related to deficiencies in the simulation of the extinction by aerosols. As a remedy, a fast yet highly accurate analytical parameterization of the aerosol transmittance is introduced here. A performance evaluation using worldwide ground observations of aerosol Angstrom's parameters has shown that, conversely to the existing parameterizations, the new one performs consistently under all possible aerosol conditions with virtually no bias and no dispersion. At seven arid locations, in particular, it is found that, when the new aerosol parameterization is used, the direct normal irradiance (DNI) predictions of the reviewed clear-sky irradiance models are comparable to those made with the high-performance RRTMG physical model. This fact proves that the deficiencies in the simulation of aerosol extinction produce most of the DNI biases of clear-sky solar irradiance models in arid regions. Thus, the performance differences found in such models in previous benchmarking studies can vanish by using a better parameterization of the aerosol transmittance such as the one proposed here.