Update 20160803.
Extent: -149.0 (-138k vs 2015, -686k vs 2014, -581k vs 2013, +69k vs 2012)
Area: -118.0 (-185k vs 2015, -998k vs 2014, -788k vs 2013, -240k vs 2012)
Wipneus, according to those numbers, the sea ice area in 2015 was 55 thousand km2 below 2012 at this date. As discussed earlier in this thread, your numbers for 2012 are calculated from UH SSMIS data whereas the numbers for 2013-2016 are from UH AMSR2 data. The graphs you posted a few weeks ago (in Reply #2384) suggest that this could affect the calculation of sea ice area by about 0.3 million km2 or so at this time of year (look at the data for August/September 2013 in that graph). So this could affect the ranking of 2012 vs. the other years in your daily reports?
It is more like 230k, which is really not bad for two different area calculations. But yes those are the margins of error that are inherent in this game.
You could add 230k to the difference numbers of 2012, and make an argument that that would be a fairer game. I hesitate to do that, 2012 may not be like 2013 or 2014, and just adding an arbitrary number without much understanding is not much to my liking.
Moreover, looking at the compactness graph in the top post of this thread:
https://sites.google.com/site/arctischepinguin/home/amsr2/grf/amsr2-compact-compare.png
the calculations with Uni Hamburg data in this graph (blue and black curves) suggest that compactness in 2012 was substantially higher than in 2015 and 2016 at this time of year. Whereas the calculations with JAXA data (green curves) suggest the opposite: compactness in 2012 was lower than in 2015 and 2016 at this time of year. Which is more realistic?
I do not really have an answer, but:
About using the SSMIS data, since both extent and area of that calculation have similar and same sign offsets, you could expect compactness (the ratio of area and extent) to be quite close and not likely to cause this interesting observation.
In the second place looking at this graph you should realize that there are bigger differences between the data sets (NSIDC, Jaxa and Uni Hamburg), than between different years. You can make a point that the three compactness groups are different any way.
Another thought is that this graph is probably too simplified by taking the whole arctic number. In summer many regions that have nearly melted out, the ice there is, has a high proportion of marginal ice. There is a rather ugly
regional compactness graph, that makes the chaotic behavior of such marginal ice compactness clear. Unfortunately I had to leave Jaxa out to keep that graph readable.