Because the surface temperature anomaly being plotted was higher at the beginning of 2007 than it was in 1998, as shows up in data sets like NOAA and GISS, as well as MSU for the lower troposphere.
Most sources talking about how cold 2007 was seem to be breathlessly quoting Motl on the MSU data, which do show it as colder but only for higher atmospheric layers. Needless to say that is not mean the graphed surface data is wrong. And it does not even mean the upper layers were particularly cold on a larger scale, because here century is really only a puffy term for the last decade.
Maybe instead of assuming I don’t understand what my graph shows, you should question whether your sources are giving you an honest picture, because this isn’t the first time they have misled you.
