As Jim Pettit has already pointed out, the "weird" thing about the IJIS figures is not so much the date stamp, but the anomalous timing associated with the appearance of the figures.
Most of the organisations that provide this data do so with a one-day lag between the value of the date stamp and the day on which the data actually appears. For example, today is the 29th, and the NSIDC data (which has just appeared in the last few hours) is dated the 28th Dec. As you all know, this is the
de facto standard we expect from the IJIS data.
However, this is a convention. One could just as easily set the date stamp to the date on which the data is
(or would normally be) posted online. This, in fact, is the technique which the University of Bremen adopts - their graphs are set to show the date when the data was updated. The graph given in the link below should demonstrate the point...
http://www.iup.uni-bremen.de:8084/amsr2/extent_n_running_mean_amsr2_regular.pngGoing back to the IJIS data, the starting point is to remember that JST is 9 hours ahead of GMT. (In Japan. there is no daylight saving offset to worry about.) Therefore, as it states on the ADS website, the target time for updating is 03:00 GMT, or 12:00 JST - i.e. midday.
Most people here will know that the IJIS data is actually a rolling two day average, and it is possible that this is the key to the quandary.
Perhaps there was some reason that the ADS people realised that the "normal" update scheduled to be posted on the 29th could not be accomplished. The ADS team might then have done something along the lines of the following...
Actual date 28th - Date stamp applied 27th - equates to average over 27th & 26th
Actual date 28th - Date stamp applied 28th - equates to single day value of 27th
I am not stating that this is an accurate representation of what happened; rather that this represents merely one of a variety of possible scenarios.