Reanalyses may have limitations in representing the solar energy disposition of the Earth–atmosphere system
Abstract. Reanalysis data, including radiation data, are widely used in weather, climate, ecology, and water-cycle research. However, comprehensive evaluations of reanalysis solar energy disposition (SED) within the Earth–atmosphere system remain scarce. Combining collocated satellite and ground observations, this study evaluates the SED of three current representative reanalyses, ERA5 from ECMWF, CFSv2 from NCEP, and MERRA2 from NASA. The results demonstrate: 1) Among the three reanalyses, ERA5 exhibits the highest accuracy and most consistent performance across stations, while CFSv2 performs slightly better than MERRA2, which displays a pronounced systematic bias; 2) All three reanalyses demonstrate limited skill in representing the variability of atmospheric absorptance, explaining only a small fraction of the observed variance (the R² values of atmospheric absorptance between ERA5, CFSv2, and MERRA2 and observations are only 0.14, 0.12, and 0.14, respectively); 3) There appears to be a systematic bias in the three reanalyses, namely that atmospheric absorptance is substantially underestimated and surface absorptance is substantially overestimated, and this discrepancy is especially evident under overcast conditions. Our evaluations reveal that current reanalysis products may still have two inherent limitations in reproducing atmospheric absorptance, which may further affect the accuracies of the other two SED components.