Many methods exist for calculating and predicting solar radiation. These can be divided into three board categories, (i) the observational or empirical methods which use air temperatures, rainfall and sunshine hours measured over a period of years or other which use the earth sun geometric relationships and the location on the earth to modify these observations; (ii) the linked or artificial intelligent (AI) methods improve observations methods by interlinking observations over a geographic area or to other weather parameters so that the predictions are based on several linked observations and topographic relationships for a more reliable forecast; and (iii) the dynamic or modelling approach is mostly used in meteorology or atmospheric sciences, which uses measured variables such as water vapour, and ozone thickness and models attenuation, scattering and absorption of solar radiation in the atmosphere over the earth. Calculation and/or prediction of the amount of solar energy a location receives are useful for many different engineering applications. For example, for agricultural applications it indicates growth rates, for infrastructure, transport and manufacturing applications it dictates material selection and frequency of maintenance, etc. This paper explores the above mentioned three methods with a view to their engineeringapplications contrasting the different accuracies to the varying needs of the renewable energy sector. Problematic issues such as computation cost, the availability of solar radiation and what dictates the quality of measurement for the solar energy sector are also discussed.