Comparison between solar radiation models based on cloud information

Simple radiation estimation models using few input parameters are often used in the applications requiring rough estimations of solar horizontal radiation. Cloud cover information is widely available for multiple regional and national weather stations. Kasten and Czeplak (1980) as well as Lam and Li (1998) have proposed cloud-based models for the estimation of global and diffuse horizontal irradiation. Muneer and Gul (2000) have analyzed solar radiation estimation models based on meteorological data and modified the Kasten-Czeplak's models to allow for local coefficients to be used. In this study, the Lam-Li model has been modified to allow for local coefficients, and three new models, which are primarily based on modifications to the Kasten-Czeplak model, have been proposed. To compare the validity of the above models, statistical indicators have been presently developed for assessing the accuracy of estimation of both global and diffuse irradiation. The slope of the bestfit line and the coefficient of determination between I-G,I- measured and I-G,I- calculated and I-D,I- measured and I-D,I- calculated, the percentage error, mean bias error and root mean square error for both global and diffuse horizontal irradiation were calculated. In addition, the kurtosis and skewness of error histograms were also obtained. These parameters were used to obtain the statistical indicators for the models under validation. It was concluded that although the proposed models outperformed the original KastenCzeplak and Lam-Li models, they performed on a par with the simpler Kasten-Czeplak model with modified local coefficients.