A-Team, ESA has a high resolution version of that figure: http://spaceinimages.esa.int/Images/2013/02/Sea_ice_thickness

I believe we have just seen the largest ever single day drop in Global CT SIA, -561k km2: 2012.9069 -0.3898046 20.2006760 20.5904808 2012.9095 -0.8500094 19.6394539 20.4894638 It was a combination of -377k drop in the south and -185k drop in the north (in November!). Even separately those numbers would be quite exceptional though possibly not unprecedented. The previous record is -523k from January 2008. There is also -1.2M in December 1987 but that is an obvious typo, sensor glitch or other error.

P-maker That theory about the polynyas formed by ocean currents comes from the cited paper by Bindschadler and others. There is a non-paywall copy here: http://www.igsoc.org:8080/journal/57/204/j10j169.pdf

Chris R, The Francis & Hunter paper you cite finds a correlation between atmospheric humidity, downwelling infra-red and ice loss, but they seem to completely ignore latent heat flux between atmosphere and ice/ocean. Latent heat flux is certainly strongly correlated with humidity; it could even change direction. Do you think latent heat flux could have a significant role here?

There are two obvious negative feedbacks: As the ice extent shrinks, more of the extra heat in the arctic goes into warming the waters rather than melting the ice. Note that this is a negative feedback with respect to ice loss but not with respect to arctic warming. As the arctic warms, the temperature difference between the arctic and lower latitudes shrinks. This tends to reduce the net heat flux into the arctic. Note again that this is a negative feedback with respect to arctic warming but not with respect to global warming. These are real negative feedbacks but I'm not sure about their timing and strength. They might be overwhelmed by other effects including new kinds of positive feedbacks like storm damage to thinning ice, but I would not be surprised to see some kind of a Gompertz like tail.

The multi-year ice video in the NOAA ClimateWatch Magazine article is fascinating to watch in slow motion. Three distinct processes behind the disappearance of MYI can be observed: Export through Fram Strait. This has been going on for a long time, I guess, so it alone does not explain the loss of MYI. Increased melting of FYI. There isn't enough FYI anymore at the end of summer to replenish the flushed out MYI. Increased melting of MYI in the Beaufort Gyre, first in the Chukchi and Beaufort Seas and later in the East Siberian Sea. This vipes out the Pacific side MYI that is immune to the lure of Fram Strait.

Nightvid Cole, Because extent is not constant, your equation needs to use the proper product rule: dV = E dT + T dE You are rigth. My equation was badly formulated. What I meant was dV = E cH where H is the average net heat flux per square kilometer and c is some constant (with appropriate units). I used dT as a proxy for cH but that is not correct as you point out. The rest of my argument still holds. H is not growing very fast anymore while E is starting to fall rapidly leading to slower ice loss. In theory.

"Keep in mind that as surface area decreases, more thickness has to be lost to get the same volume drop, so it becomes harder." I've been thinking about this too and came up with this equation: dV = E x dT where dV = volume loss rate, the (average) annual loss of ice volume from September to September. E = sea ice extent in September. dT = thickness loss rate, the (average) annual loss of ice thickness from September to September. For example the equation for 80s is: 75 km3/a = 7.5M km2 x 10 mm/a and for the decade 2002-2012: 750 km3/a = 5M km2 x 150 mm/a (These are rough numbers obtained by eyeballing some graphs.) The rationale for the equation is that any additional heat input (such as absorbed insolation) to the September extend area is almost completely spent on melting ice that would otherwise survive through the melt season. Additional heat input elsewhere ends up warming the ocean waters instead. Up to now, the equation has been dominated by the dramatic increase in the thickness loss rate, which can be explained by two factors. First, when you start low, near zero, as in the 80s, a small increase in the average temperature can cause a large relative growth of the melt rate. Second, feedback effects like meltpond albedo effect and loss of multi-year ice have caused further acceleration. I'm not expecting this to continue, however. The thickness loss rate is already so high that any further growth is likely to be relatively small. In contrast, the reduction of ice extent is becoming more and more significant in the relative sense. This could mean a sigmoidal tail to the curves. The real world is not that simple of course. There are effects like ice export through Fram strait and storm damage to the thinning ice that do not fit in the equation. Also plain natural variation can muddle the picture. 2007 and 2010 saw the ice volume drop by 2500 km3 from the previous year. A similar drop next year would leave less than 1000 km3 of ice. Thus, while there may be theoretical reasons to expect a sigmoidal tail, the ice might not survive long enough for us to see it through the noise.

Jim, "(I think the state of knowledge is still as represented, but I'm not positive.)" Found a couple of recent papers: http://dx.doi.org/10.5194/cp-7-603-2011 http://dx.doi.org/10.5194/cpd-8-2969-2012 It seems that problems with the proxy data explain at least part of the discrepancy but the models are not in the clear either.