The difference is in the 2007 data
CT had much less ice extent than NSIDC in mid-August last year, while this year's data are very similar.
Compare http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/ARCHIVE/20070812.jpg and ftp://sidads.colorado.edu/pub/DATASETS/seaice/polar-stereo/nasateam/final-gsfc/browse/north/daily/2007/nt_20070812_f13_v01_n.png
I made an overlay of both, as far as possible due to different map projections: http://i36.tinypic.com/nxjteb.png
MODIS image of the eastern Beaufort Sea at that day: http://rapidfire.sci.gsfc.nasa.gov/realtime/single.php?T072240405 (cloudy elsewhere).
CT images had much fluctuations from day to day at that time; NSIDC images were much more stable. The increase between 2007-08-12 and 2008-08-11 in extent based on CT images is indeed somewhere in the range 20-40 %, but comparing single days doesn't make much sense with that much daily fluctuation (apart from the questionable method of pixel counting in anti-aliased and distorted images).