Calibration and evaluation of sunshine-based empirical models for estimating daily solar radiation in Iran

• Published in: Sustainable energy technologies and assessments 2020-12, Vol.42, p.100855, Article 100855
• Main Authors: Naserpour, Somayeh, Zolfaghari, Hasan, Zeaiean Firouzabadi, Parviz
• Format: Article
• Language: English
• Subjects: Calibration; Empirical models; Iran; Linear and nonlinear regression models; Solar radiation
• ISSN: 2213-1388
• DOI: 10.1016/j.seta.2020.100855

•21 sunshine-based empirical models were calibrated and evaluated.•The constant coefficients of models have estimated by the best data fit.•M 4 &12 in most regions show the highest correlation with actual radiation values.•M 15 has well performed in the climatic conditions across the whole country.•For all stations the highest radiation was in June and the lowest in December. This research aimed to find a proper model to estimate daily solar radiation in terms of specific climate conditions of 51 weather stations over Iran. Twenty-one sunshine-based empirical models were calibrated and evaluated. Two different quality control methods were used to reject data error in daily solar radiation dataset collected from 2007 to 2017. The empirical constant coefficients were calculated using linear and nonlinear regression models. Statistical error comparisons between observation and models-based estimates were performed via R2, RMSE, RRMSE, MAPE, MBE and MAE measures. Finally, following the models-based estimates, monthly solar radiation maps were created for the entire country using ordinary Kriging interpolation method. The results revealed that for the largest part of the regions, the highest correlation between Observation and Models-Based Estimates was detected while implementing regression-based models (models 4 and 12). However, the highest correlation between the values of the measured radiation and the models-based estimates over the entire country belonged to the linear model 15. Moreover, there was an opposite relationship between “a” and “b” coefficients in model 15. The results of models 4 and 12 show that these models can perform well for estimating solar radiation in the “C” and “F” climate zones, respectively.