New decomposition models to estimate hourly global solar radiation from the daily value

Abstract

Daily global solar radiation decomposition models (DGSRD models) are important to solar photovoltaic, solar thermal utilization and building energy assessment. The existing DGSRD models are analyzed with hourly global solar radiation data measured from January 2009 to December 2011 on Jiading Campus, Tongji University. The results show the existing models established in the form of a Gaussian equation are comparatively accurate, followed by models adjusted or modified from the Whillier – Liu & Jordan models, and the Newell model is of the lowest accuracy. Among the existing models, the CPRG model is the most accurate one. Then four new decomposition models of daily global solar radiation (DGSRD model 1 to DGSRD model 4) are established based on consideration of the solar altitude angle, the solar azimuth angle, the solar hour angle, clearness index and temperature. Based on the analysis of correlation coefficient, relative standard error, mean bias error, mean absolute bias error, root mean square error, Nash–Sutcliffe Equation and t-statistic (t-stat), it is demonstrated that new DGSRD models (DGSRD model 1, DGSRD model 2 and DGSRD model 4) are more accurate than those existing ones, in which the DGSRD model 4 is the most accurate one.