Improved Global Irradiance Decomposition by Sky Condition Classification from Measured Spectral Clearness Indices

This manuscript describes a decomposition algorithm for deriving the broadband direct normal (DNI) and diffuse horizontal (DHI) irradiances from one-minute spectral global horizontal irradiance measurements performed by the SolarSIM-G. The algorithm was calibrated and validated at four stations across Canada (Ottawa, Varennes, Egbert, and Devon) and at one station in China (Xianghe), which cumulatively represent a seven-year data set. For the DNI estimation, the root mean square error (RMSE) ranged from 26 W/m
²
to 48 W/m2, while the largest mean bias error (MBE) was 4 W/m
²
. For the DHI estimation, the RMSE ranged from 14 W/m
²
to 27 W/m
²
, while the largest MBE was 3 W/m
²
. In addition, the integrated DNI and DHI errors for each station were less than 1% and 2%, respectively. The described method is an alternative to other measurement techniques for obtaining all three irradiance components.