Comparative study of mathematical models in estimating solar irradiance for Australia

Hourly diffuse and direct solar irradiance data are required for weather files used in building energy simulation as well as Photovoltaic and solar thermal calculations. Access to up to date hourly observation data or satellite derived data for these parameters is currently only available for a selection of the Australian Bureau of Meteorology measurement locations. This study aims to investigate the accuracy of the methods used to estimate solar irradiance data from meteorological observations either in the absence of observed irradiance data or when irradiance observations are limited to global irradiance. In particular the focus of this paper is to investigate the accuracy of the process of coupling together global and direct/diffuse models. Five global and four diffuse/direct irradiance models are presented and compared to experimental data for four locations in Australia. In the case where experimental global irradiance data is available, values are used as input into various models to obtain diffuse and direct irradiance. In the absence of experimental irradiance data, the approach taken is to estimate global irradiance with a separate model and then feed these values into the diffuse/direct models. The errors associated with both of these approaches are investigated by comparing the modelled diffuse and direct irradiance values with known experimental data for four Australian locations over a period of a number of years. This study indicates that no single diffuse/direct irradiance model consistently outperformed the other models at estimating diffuse and direct irradiance whilst the Zhang and Huang global irradiance model with coefficients from Seo and Huang achieved the best estimates of global irradiance for the locations investigated. For the approach where global irradiance is estimated from a model, the resulting direct and diffuse data was found to differ significantly from the experimental data. The results indicate that the individual models that achieved the best estimates of global irradiance did not achieve the best estimates of diffuse and direct irradiance when coupled with a diffuse/direct model.

► Comparison of five global and four diffuse and direct irradiance models.
► Results presented combining global and diffuse irradiance models.
► Model accuracy and suitability dependent on location and frequency of cloud cover.