Estimating the global solar irradiation and optimizing the error estimates under Algerian desert climate

Accurate predictions of the solar resource are essential tools for the design, performance analysis, and economic evaluation of various solar energy projects. This paper is proposed in an attempt to overcome the lack of solar radiation data and predicting models in Algeria. New and readjusted models have been developed to estimate the daily and monthly mean daily global solar irradiations over Algerian Big South, based on measured ground data of air temperature and sunshine hours, as well as the day/month number. After a careful evaluation of the models’ performances using the standard K-fold cross-validation method, it was found that only the sunshine-based models can offer excellent estimations in the daily time scale, with validation RMSEs in the range of 1.470–2.425 MJ/m2 day. Meanwhile, all types of models have superior estimations of the monthly mean global irradiation (RRMSE less than 10% in most cases). The merits of using the K-fold cross-validation method in optimizing the error estimates, stabilizing the performances, as well as selecting the appropriate models, have been demonstrated.