National-scale development and calibration of empirical models for predicting daily global solar radiation in China

Accurate global solar radiation (R-s) information is pivotal to the design and management of solar energy systems. Nevertheless, the expensive devices for R-s measurements make R-s data always unavailable in many regions around the world. The empirical models that predict R-s using other widely available climatic variables are feasible alternatives when R-s measurements are unavailable. However, the parameters of empirical models are site-specific and always need local calibration. In this context, the present study firstly developed a novel empirical model for accurately predicting R-s at the national scale in China, and then compared the new model with nineteen locally calibrated empirical models that have been largely reported in prior studies, including seven sunshine-based, nine temperature-based, and three complex empirical models. Daily R-s and other meteorological data during 1994-2016 from 96 weather stations in China were used to calibrate/develop and assess the models. The results showed that the newly proposed C4 model generally offered the best prediction accuracy among the models, with average MAE of 1.69 MJ m(-2) d(-1), RRMSE of 16.2% and NS of 0.895, which can be recommended as the optimal model for predicting R-s in China. The models reviewed and developed in this study improved the prediction accuracy of R-s which can provide crucial information for the design and implementation of solar photovoltaic and thermal systems.