In this study, measured data of solar radiation were utilized to develop sixteen empirical models for estimation of monthly mean diffuse solar radiation with single input predictor as clearness index. Global and diffuse solar radiations were measured for three complete years in ‘Humid subtropical climatic region’ of India. Equipment consisting of secondary standard pyranometers and continuous datalogging facility were used to observe solar radiation at ground level in the city of Aligarh (27.89°N, 78.08°E). Two categories of models were defined (i.e. diffuse fraction and diffusion coefficient) with eight models under each category. Total solar radiation dataset was divided into two parts where ‘Training dataset’ was used to develop the models while the ‘Validation dataset’ was used to test the models. Accuracy of models was analyzed in terms of some commonly used statistical indicators. Developed models were also compared with well-established models from literature. Models were arranged in order of suitability of estimation using Global Performance Indicator within the respective category as well as among the overall group of developed models. Reasonable agreement was found between estimated values from the developed models and measured data. Developed models can conveniently be utilized to evaluate diffuse solar radiation in the region considered.
Read full abstract