Estimation of surface longwave radiation components from ground-based historical net radiation and weather data

A methodology for estimating ground upwelling, clear-sky and cloud downwelling longwave radiations (L-up arrow, L-sky(down arrow), and L-cld(down arrow)) and net shortwave radiation (S-n) at 30-min temporal scales based on long-term ground-based net radiations and meteorological observations is described. Components of surface radiation can be estimated from empirical models, cloud radiation models, and remote sensing observations. The proposed method combines the local calibration of empirical models and the radiative energy balance method to obtain the dual-directional, dual-spectral components of the surface radiation for the offline land surface process modeling and ecosystem study. By extracting information of radiation components from long-term net radiation and concurrent weather data, the utility of tower net radiation observations is maximized. Four test sites with multiyears' radiation records were used to evaluate the method. The results show that when compared with the results of empirical models using default parameters the proposed method is able to produce more accurate estimates of longwave surface components (L-g(up arrow), L-sky(down arrow), L-cld(down arrow)) and net shortwave radiation (Sn). Overall, the estimated and observed surface radiation components show high correlations (>0.90), high index of agreement (>0.89), and low errors (root mean square error <30 W m(-2) and bias < 11 W m(-2)) at all of the test sites. The advantage of this scheme is that the refinement is achieved using the information from the historical net radiation and weather data at each observation site without additional measurements. This method is applicable for many existing observation sites worldwide which have long-term (at least 1 year) continuous net radiation records.