Comparative Analysis of Estimation Models for Monthly Average Hourly Solar Radiation

Accurate hourly or sub-hourly monthly average solar radiation estimation models gain importance in scenarios of increasing penetration of grid connected photovoltaic systems, which is a key energy generation technology for the transition to a low-carbon society. Initially, this paper reviews the main models in literature and proposes a standardized nomenclature. The models are analyzed in relation to implementation, expected behavior and environmental and locational factors. Their performances are verified and compared by means of statistical analysis using data from forty Brazilian meteorological stations. A new set of three statistical metrics that allow decoupling the effect of local asymmetry between morning and afternoon on models’ errors is proposed, leading to the calculation of the theoretical maximum mitigable error. To improve models’ performance, replacement of the hourly fraction concept by a new one, the instantaneous fraction, and allowing the synthesis of solar radiation in sub-hourly intervals are recommended. It is concluded that asymmetry is an important source of errors in several situations. Besides, no existing model has satisfactory results for all monthly climatic conditions, and new, more accurate and widely validated models are welcome.