Estimation of high-resolution solar irradiance data using optimized semi-empirical satellite method and GOES-16 imagery

Semi-empirical satellite method is widely used in estimating global horizontal irradiance (GHI), where various clear-sky models, cloud index (CI) and clear-sky index (CSI) derivation methods are available. This study aims to optimize the semi-empirical satellite model for 5-minute GHI estimation from four aspects: satellite-bands, CI and CSI derivation methods, and clear-sky models. The results show that it achieves better GHI estimates using the blue band, CI derived from monthly fixed upper and lower bounds, and a piecewise CI-to-CSI function. There is no significant difference in all-sky GHI estimation for the clear-sky models regarding normalized root mean squared error (nRMSE, 25.19%–25.53%), which is comparable with the referenced physical model. Clouds cause the largest uncertainty, where the nRMSE is in the range of 37.60%–38.36% in cloudy days and 31.12%–31.54% in cloudy periods. In the application of semi-empirical method with different clear-sky models, Ineichen–Perez has the highest bias of -4.62% in clear days and -3.93% in cloudless periods. REST2 outperforms McClear with slightly lower nRMSE and normalized mean bias error (nMBE) under all sky conditions. McClear is recommended due to its global availability. Modified Ineichen–Perez produces the lowest nRMSE and nMBE using clear-sky GHI as the GHI estimates for clear periods, therefore has the potential for improvements in physical methods.