Figure 15 in the paper presents their model's results, side-by-side with satellite measurements. To the naked eye, they look good. However, when they are plotted against each other, they do not match. As their model is meant to study spectral absorption, you might think their model would match the satellite data in the "spectral window" where no absorption occurs. It does not. Their model must be scaled by +5.6%, -3.9%, and -12.5% for the Sahara, Mediterranean, and Antarctic datasets, respectively. Even after adjusting the model to match the satellite data, the model overestimates the amount of outgoing radiation. This infers that the Earth is staying cool, when in fact it is retaining heat. The discrepancy is on the order of 3-5% globally, which is quite large.
In answering this question, I have used the data presented in the paper itself. As the authors did not provide the raw data, I used software to digitize the graphical data. My results may be off by a decimal place, but they are in the ballpark.