This study analyzes the downwelling longwave radiation (DLW) over the Greenland Ice Sheet (GrIS) using surface-based observations from Summit Station (72 degrees N, 38 degrees W; 3210m) and the European Centre for Medium-Range Weather Forecasts Interim Reanalysis (ERA-Interim) DLW fields. Since surface-based observations are sparse in the Arctic, the accuracy of including reanalyses for spatial context is assessed. First, the DLW at Summit is reported, including the significant time scales of variability using time-frequency decomposition (wavelet analysis). A new method for evaluating reanalyses is then introduced that also uses wavelet analysis. ERA-Interim DLW performs reasonably well at Summit, but because it includes too many thin clouds and too few thick clouds, it is biased low overall. The correlation between the observations and ERA-Interim drops from r(2)>0.8 to near 0 for time series reconstructed from time scales less than similar to 4days. These low correlations and additional analyses suggest that the spatial resolution of the data sets is a factor in representing variability on short time scales. The bias is low across all time scales and is thus likely tied to cloud generation processes in the model rather than the spatial representation of the atmosphere across the GrIS. The exception is autumn, when ERA-Interim overestimates the influence of clouds at time scales of 1 and 4weeks. The spatial distribution of cloud influence on the DLW across the GrIS indicates that Summit is located in a transition zone with respect to cloud properties. The gradient across this transition zone is steepest near Summit in autumn, so the spatial characteristics of the atmosphere near Summit may contribute to the ERA-Interim bias during this time.