Modelling the global solar radiation climate of Mauritius using regression techniques

The tropical island of Mauritius (20.3°S, 57.6°E), situated in the southwestern Indian Ocean, is blessed with abundant sunshine throughout the year. In this study, eleven regression models grouped into three main categories: sunshine-based, temperature-based and hybrid-parameter-based are investigated using twenty-nine years meteorological data which include sunshine hours, temperature and relative humidity for fifteen stations on the island. Calibration of these models in the varying climatic regimes of the island is performed using global solar irradiation measurements recorded at the fifteen stations. An additional model, Sayigh Universal Formula, is modified through the implementation of a relative humidity factor trend specific to Mauritius and applicable to all regions on the island. The prediction capabilities of all twelve models are determined using statistical evaluation indicators and the better performance of the Sayigh Universal Formula acclimatized to Mauritius is revealed. Spatially clustered cloud cover zones are found to influence significantly the spatial distribution of global solar irradiation on a horizontal surface on the island, which varies from a maximum value of 22.5 MJ/m2day to a minimum of 9.5 MJ/m2day throughout the year giving an average of about 16 MJ/m2day.