Abstract
Estimations of clear-sky Global Horizontal Irradiance (GHI) are essential to assess the solar resources. These clear-sky models are the basis when studying any solar project. However, the accuracy of these models differs significantly, that is mostly related to the geographical location of the site under test, as well as the input variables to the model and the goodness of the model itself. In this work, we focus on studying different clear-sky models, where we examine 25 popular clear-sky GHI models using data from two sites in Riyadh, Saudi Arabia. A dataset that covers the period of 7 years was used, with an hourly readings. Since the validation of GHI models requires comparison of the model results with measured values during clear-sky periods, Polo's algorithm was used to determine clear days. Clear-sky models' performance is evaluated using the Combined Performance Index (CPI), which combines Root Mean Square Error (RMSE), Kolmogorov-Smirnov test Integral (KSI) and OVER parameter into single statistical indicator. Results show that 7 models have less than 50% Combined Performance Index (CPI), where MAC, and REST2 clear-sky models has the best performance with less than 25% CPI. Overall, the top performing models are the ones that include the climatic variables, which indicate the importance of considering the climatic and geographical conditions when selecting a model.