Also see this comparison of collapsing ice sheets with avalanches:
http://johnenglander.net/sea-level-rise-blog/inability-predict-slr-similar-avalanche-problem
I hope I misunderstood this article... He seems to tell that SLR from ice sheet collapse cannot be predicted by comparing this to the problem of predicting
when an avalanche or earth quake gets loose. That looks totally wrong to me: The collapse will not come sudden, like a house collapsing due to an earth quake. No, the collapse is already happening! (E.g. Rignot et al found WAIS collapse now is even irreversible). But it is happening in slow motion.
The avalanche metaphor seems quite apt, but only when taken in slow motion. (I'm no ice sheet expert and wanted to ask for this today somewhere on the forum - so I searched for the key word "exponential" and thus came here.
)
The ice sheet "avalanche" is already rolling upon us, meanwhile, accumulating in mass, growing more exponentially than linearly. (+). Taking a conservative estimate of 10y doubling time of melt rate and taking just Greenland, then we get 3.5-7 meters SLR by 2100, as Hansen said long time ago. (Now imagine that it looks more like 5y doubling time - the avalanche then growing monstrously bigger and faster. I hear it roaring...)
Problem with extrapolating exponential trends is of course the growing error margin (E.g.: Is it half of Greenland by 2100 or all of it?). But that doesn't mean prediction is impossible - you just have to get rid of the common image of a linear scale of numbers. (Maybe play with an ancient mechanical slide rule calculator. Every pre-calc math student should have one.)
Albert Bartlett's famous dictum:
The greatest shortcoming of the human race is our inability to understand the exponential function.
(Well, methinks that's due to the cultural dominance of left brain hemisphere thinking, which only can do linear bean counting (however complex it may be) but fails at holistic systems grasp. Grasping exponential stuff (and its vagaries in a finite real world) is more for the right hemisphere. But that's for a different thread...)
--------------------
(+) Edit, P.S.:
There are general systems thinking reasons for preferring exponential over linear. (A bit more detail in my 2009 comment
here.) Me dunno know more. And Hansen said the same.
Meanwhile, my impression is that it's almost consensus amongst ice sheet experts to assume exponential decay (and monstrous SLR as a corollary) - drawing from their
expert intuition about the concrete physical system at hand.