Let's return to the central issues...
1) When the Arctic will go ice free
2) How that is likely to play out from first ice free day, to week, to month, to season, to year
3) What the consequences are of that, and hence why we should care
1) the trends in ice extent, ice area and ice volume are all headed to the same outcome, zero ice. Each points to a somewhat different potential date for that. The differences in those dates, though important from a human perspective in a single human lifetime, are essentially identical in geologic terms, and virtually identical in the lifetime of civilizations or nations.
The most likely correct projection is the limiting projection based on the full suite of projections, not the average, not the last, but the first. And that is based on volume. The inherent oscillatory nature of the many linked earth and solar systems creates a form of variation that looks like and can useful be treated similar to randomness. And it has randomness in it. But it isn't truly random in the large scale.
That said, the outer bounds of the error band on projecting forward on ice volume suggest that we have already entered the outermost likelihood for an ice free summer day. Clearly this year won't be it. Next year could be. But most likely that won't be for a few years.
On the other end, the high band, we almost certainly will see it before 2030 even under the most unlikely combination of events. As a result, the first ice free day in September will almost certainly occur between 2022 and 2028.
2) with the progressive loss of ice cover, warming of the ice free ocean, thinning of the ice cover, failure of the tundra and clathrates, combined with mans continued and accelerating release of global warming gases, the lengths of time that the Arctic is essentially ice free will grow longer. There will be oscillation with temporary retreats, and with shocking extensions. The trend will remain for longer and longer ice free periods. That will happen quickly, even in human terms.
3) as that happens, the downwelling driving forces on both the ocean, driving the Atlantic and Pacific oceanic circulations will progressively grow weaker, and the down falling driving force for the atmosphere will simultaneously decline with it, and with that the motive forces for atmospheric circulation of the polar cell will decline.
As the oceanic driving forces collapse a whole suite of interlocking circulations will lose their motive force. New balances will come into play. The oceanic circulations will perhaps stall, and in some areas new broader slower circulations driven by corriolis forces and topography will take over. Areas will go anoxic. Species will move with the temperature and flow. Many will die.
As the atmospheric driving forces fail, the heat balance will shift. The tropopause will rise. The polar circulation will slow and become more chaotic before too be driven by lesser circulations and forces. As the polar cell fails, so too will the driving forces between the Ferrell and polar cells weaken and fail, then those between the Ferrell and Hadley cells. In time, those too will be overridden by other forces.
With an increased tropopause, single cell circulation becomes possible, though moving at slower speeds allowing drag to counter corriolis forces that would otherwise truncate the circulation. Exactly what happens with this is unknown and is a key question related to how the atmosphere circulates on Venus, and how it circulated on Earth during equable climate periods.
The oceanic and atmospheric circulations are however also interdependent based both on flow interactions and based on heat. With dramatic shifts in flow and consequent large shifts in heat balance, moisture shifts, clouds and the like, the problem is extraordinarily difficult to sort out.
That it will shift is certain.
As has already been noted, we are already seeing dramatic shifts in all of these, with dramatic consequences. However, the largest differences will no doubt come when the relative balance between the various forces reach near parity. At that point, if we had a non dimensional analysis to guide us, we might (and only might) have a better idea about how the transitions will occur, and precisely when we might expect hysteretic sorts of state change.
I haven't found a non dimensional analysis of the coupled ocean, air, ice thermodynamic system using the Buckingham Pi method that might aid there. If anyone does, that might be quite useful. It should tell us what the key dimensionless parameters are to monitor (essentially the ratios of various forces that drive the system as a whole).
What we can be certain of is that the Earth is a heat engine. During periods such as our recent several millions of years where we have ice at the poles, the heat differential between these and the solar inputs (dominant at the equator) act to stabilize the system like a giant engine. The ice acts as a huge buffer or battery holding the system in a sort of equilibrium. That oscillates annually and at longer periods. Still it is a buffer. With the loss of that buffer, the system loses its governor. It then is likely to change quite quickly to an alternate stable system governed by other dynamics. That is when we will,see and experience truly abrupt climate change. No one will need convincing then that it is real. But, no doubt, many will still need convincing that we are at fault, and that we need to urgently act.
That we don't know those dynamics sufficiently well to model them successfully is particularly troubling. That we know from geologic records just how different that system is is even more troubling. But, and this is especially important, people lose sight of the importance of the rate of change in converting from one to state to another. Prior geologic analogies seem tame and slow by comparison to our current predicament. And this may be why a period of ice free Arctic in and transition period between ice ages could exist without completely upending the system. Even then, the dynamics are such that the conditions must have been radically different from what we are acuustomed to.
In our case though, we don't have slow changes at work. Our case is more akin to a fully loaded 18 wheeler racing down a 12% grade, burning out its breaks and bashing through the guardrail into open air several thousand feet above the canyon floor. You might as well decide to enjoy the ever so brief ride, as no amount of steering or cranking on the breaks means anything at that point.
But in our analogy we are still on the road. We've begun to lose traction with the highway, the breaks are all but gone and the steering isn't working. Worse, we are making our decisions by committee with a crew in the cab that is, shall we say, less than up to the task.
We are in the ever so brief period before calamity where we cannot be quite certain whether we are going to inevitably go through the guard rail and plummet to our certain death, or miraculously gain the ever so small bit of control that allows us to steer onto the truck runaway ramp. Sure, it's going to rip the wheels off and all but destroy the rig, but at least we get to recover from it.
Now, if we can just get all of the monkeys in the cab to come to agreement that we need to act, and act together, maybe we might just barely survive this yet. But first we have to get them to stop biting each other and throwing their poo.
Sam