Abstract
Several empirical models of hourly ultraviolet total radiation (UVT) have been proposed in this study. Measurements of UVT radiation, 290–385 nm, have been recorded at ground level from February 2001 to June 2008 in Valladolid, Spain (latitude 41°40′N, longitude 4°50′W and 840 m a.s.l.). The empirical models have emerged due to the lack of some radiometric variables in measuring stations. Hence, good forecasts of them can be obtained from usual measures in these stations. Therefore, some advantages of the empirical models are that they allow the estimation of past missing data in the database and the forecast of future ultraviolet solar availability. In this study, reported models in the bibliography have been assessed and recalibrated. New expressions have been proposed that allow obtaining hourly values of ultraviolet radiation from global radiation measures and parameters as clearness index and relative optical air mass. The accuracy of these models has been assessed through the following statistical indices: mean bias, mean‐absolute bias and root‐mean‐square errors whose values are close to zero, below 7% and below 10%, respectively. Two new clear sky models have been used to evaluate two new parameters: ultraviolet and global cloud modification factors, which can help to understand the role of the clouds on solar radiation. The ultraviolet cloud modification factor depends on cloudiness in such a way that its value under overcast skies is half of the cloudless skies one. Exponential and potential fits are the best relationships between both cloud factors. Finally, these parameters have been used to build new UV empirical models which show low values of the statistical indices mentioned above. Copyright © 2009 Royal Meteorological Society