A statistical approach, which partitions surface net radiation measurements into up- and down-welling solar and thermal components, is presented in this study to enhance the utilization of flux tower observations for the offline land surface simulations. The method uses long-term, nighttime (solar radiation free) net radiation measurements to extract the clear-sky incident long-wave irradiance, and the “observed” clear-sky radiation data that are used for calibrating the parameters in an empirical estimation method to improve the model performance. A scheme which interpolates the “critical” cloud quantities between the early morning and the late afternoon was introduced to estimate the cloudiness variation during daytime and afterwards determine the daytime cloud downward radiations. Net solar radiation is calculated from the surface radiation budget. Comparisons were made between the estimated and the observed radiant components at four AmeriFlux sites for multiyear periods. The agreement was good with an overall root mean square error of long-wave up- and downward radiation estimations less than 27.03 Wm-2 and the correlations greater than 0.89 at the hourly scale. Considering that the land-surface net radiation is the only radiant measurement available over the majority of the field measurement sites which have been built worldwide for the studies of climate, hydrology and ecology, the developed approach can enhance the utilization of those field data.