Abstract
Modelling in ungauged catchments has always been a complicated issue due to scarcity of the requisite data. Therefore, hydrologists have to rely on indirect techniques of estimation. A novel approach is hereby presented for solar radiation estimation in ungauged catchments using readily available datasets of temperature and precipitation. The rationale is that the extraterrestrial radiation is attenuated not only by different atmospheric processes but also by weather phenomena. A comparison of four clear sky radiation models was made and the best model output together with temperature and precipitation data was analysed using the gamma test for best input combination and data length selection. This led to the development of a non-linear artificial neural network model for solar radiation estimation. Results show a good correlation between the observed and estimated values, depicting the usefulness of gamma test in model development. The study is novel in the sense that this is the first time the gamma test has been used for hourly solar radiation estimation in ungauged catchments and hence extends their solar radiation records. Moreover, the regionalisation of the proposed model to other catchments would also aid solar radiation estimation in such catchments.