Abstract
A model to predict the mean expected daily global solar radiation, H(n) on a day n, at a site with latitude φ is proposed. The model is based on two cosine functions. A regression analysis taking into account the mean measured values Hm.meas(n) obtained from SoDa database for 42 sites in the Northern Hemisphere resulted in a set of mathematical expressions of split form to predict H(n). The parameters of the two cosine model for 0o<φ<23o are obtained by regression analysis using a sum of 38 Gaussian functions, while for 23o<φ<71o the two cosine model parameters are expressed by a sum of exponential functions or the product of an exponential and a cosine function. The main equation of the model and the set of parametric expressions provide H(n) for any φ on Earth. Validation results of this model are provided along with the statistical estimators NMBE, NRMSE and tstatistic in comparison to the corresponding values from three databases of NASA, SoDa and the measured values from ground stations provided in Meteonorm.
Original language  English 

Pages (fromto)  178188 
Number of pages  11 
Journal  Renewable Energy 
Volume  91 
Early online date  26 Jan 2016 
DOIs  
Publication status  Published  Jun 2016 
Keywords
 daily solar radiation
 universal model
 prediction
Profiles

Eleni Kaplani
 School of Engineering, Mathematics and Physics  Associate Professor in Solar Energy and Photovoltaics
Person: Academic, Teaching & Research