Assessment of atmospheric emissivity models for clear-sky conditions with reanalysis data

Atmospheric longwave downward radiation (L_d) is one of the significant components of net radiation (R_n), and it drives several essential ecosystem processes. L_d can be estimated with simple empirical methods using atmospheric emissivity (ε_a) submodels. In this study, eight global models for ε_a were evaluated, and the best-performing model was calibrated on a global scale using a parametric instability analysis approach. The climatic data were obtained from a dynamically consistent scale resolution of basic atmospheric quantities and computed parameters known as NCEP/NCAR reanalysis (NNR) data. The performance model was evaluated with monthly average values from the NNR data. The Brutsaert equation demonstrated the best performance, and then it was calibrated. The seasonal global trend of the Brutsaert equation calibrated coefficient ranged between 1.2 and 1.4, and the K-means analysis identified five homogeneous zones (clusters) with similar behavior. Finally, the calibrated Brutsaert equation improved the R_n estimation, with an error reduction, at the worldwide scale, of 64%. Meanwhile, the error reduction for each cluster ranged from 18 to 77%. Hence, Brutsaert's equation coefficient should not be considered a constant value for use in ε_a estimation, nor in time or location.