At the risk of polluting this thread, and with apologies in advance for not following this thread as carefully as I should...

I've been working on a theory that degree-days are of primary importance in determining arctic ice growth and ice melt. See for example

NSIDC Ice Growth and

Modeling Glacier Melt.

Jim Hunt kindly gave me the DMI 80N temperature data since 1958. [I had to interpolate data for a small number of dates for which there was no data.] For each date, if the temperature was below 271.35 degrees kelvin (-1.8C) I counted the degrees below 271.35 as freezing degree days. Similarly, if the daily temperature was above freezing, the degrees above freezing were counted as thawing degree days. This gives the following graph:

80N Saltwater Degree DaysIn this graph, the freezing degree days is on the left and the thawing degree days in on the right. There are far more freezing days than thawing days. Ice freezes down from the bottom and builds insulation as it grows, but melts from the top.

As expected, the number of freezing degree days drops over time. Unexpectedly, the thawing degree days also drops, although slower, over time as well. I'm guessing the explanation for this is that Multiyear Ice has been melting. Multiyear ice contains less salt and melts at a higher temperature. The air temperature is driven toward the temperature of the ice. With lots of Multiyear Ice in the CAB, the average summer temperature should be close to the freezing point of fresh water, as that melts the average temperature should drop toward the freezing point of salt water.

Next, we can use degree days to estimate the amount of ice that will freeze or melt. NSIDC kindly provides a formula for ice growth at the above link: 1.33*(degree_days^0.58) gives centimeters of ice growth.

For melting, the above glacier link suggests a range of multiplicative factors for glaciers and suggests that oceans have higher multiplicative factors. I arbitrarily chose 1 cm of melt per degree day. This seems to be roughly in the right ballpark, but we may want to mentally slide the line showing melted ice up or down. We get the following graph for the amount of ice grown and melted:

80N Ice Growth and MeltThe graph of course just gives a rough estimate. As ice is pushed toward the archipelago, thinner ice may be left behind causing more ice to grow than is estimated here. We start the freezing season with quite a bit of ice in place already, so we may overestimate the amount of ice that grows. And, of course, I pretty much totally made up the amount of melt. But the graphs are roughly in the right place: the growth and melt must have been in rough equilibrium in past decades or we would have seen rapid accumulation or loss of ice.

Finally, we can add the growth and melt for each year:

80N Ice ChangeThis has a large dip in 1998, and more recent dips in 2002, 2007, 2011, and 2012. So the result seems to correlate roughly with extent graphs. (These graphs are using calendar year freezing as opposed to seasonal freezing, which might produce a slightly different picture.)

So far in 2015 (as of a few days ago, day 203), we've had 123 saltwater thawing degree days, the same as we had in 2012.

To extrapolate into the future, we would expect thawing degree days to continue to drop, and thus ice growth would continue to slowly drop. I would expect thawing degree days to level out now that most multiyear ice has been lost from the CAB. But I wouldn't expect thawing degree days to rise much unless we start to melt out the periphery of the CAB relatively early in the melting season.