I don't have access to the full paper, but my reading of the abstract is (in short) that satellite measurements of independent quantities have ratios that are only (sort of) reproduced by models that show a higher warming. This leaves two possibilities: the models are wrong or the measurements are off.