Thursday, November 14, 2013

Miscellaneous Debris, or the AR5 Estimates of SLR

Ever since this came up in a twitter exchange between Aslak Grinsted and Tamsin Edwards somewhere about Halloween,

Eli has been peeking at a couple of pages on the formers web pages fittingly enough for an ice sheet expert called miscellaneous debris.  As the bunnies will see, it really is scary stuff.

Grinsted, as many ice sheet folk, is, shall the Rabett say, not very impressed with the AR anything's estimate of sea level rise from the wastage of ice sheets, which he describes as optimistic and overconfident, and ok, the AR5 did put a caveat in in fine print, that well, if there is a collapse in Antarctica all bets are off.  Grinsted shows the various estimates of sea level equivalent rise due to the ice sheets wasting away.

Most of the labels are self explanatory, but you can get the details amongst the debris.  The ranges are the 5-95% range. the whte line the best guess.
AR5 process based model projections are much more conservative/optimistic and has much more narrow uncertainties than the ice sheet experts (Fig.1). There can be no good reason for why the AR5 authors have much greater confidence in their ability to project ice sheet loss than ice sheet experts themselves. Notably the best guess view of ice sheet experts nearly falls outside the AR5 process based range. The worst case scenario from ice sheet experts is more than 60 cm higher than the worst case from the AR5 process models.

Clearly the process based SLR projections from AR5 are over-confident and too conservative by themselves.
The ice sheet experts estimates come from Bamber and Aspinall (2013). 
A major gap in predictive capability concerning the future evolution of the ice sheets was identified in the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change. As a consequence, it has been suggested that the AR4 estimates of future sea-level rise from this source may have been underestimated. Various approaches for addressing this problem have been tried, including semi-empirical models and conceptual studies. Here, we report a formalized pooling of expert views on uncertainties in future ice-sheet contributions using a structured elicitation approach. We find that the median estimate of such contributions is 29cm—substantially larger than in the AR4—while the upper 95th percentile value is 84cm, implying a conceivable risk of a sea-level rise of greater than a metre by 2100. On the critical question of whether recent ice-sheet behaviour is due to variability in the ice sheet–climate system or reflects a long-term trend, expert opinion is shown to be both very uncertain and undecided.
Grinsted points out that in the current situation, although the uncertainty monster may not be perfectly represented in models, the experts have laid hands on the beast and they
. . . know the models and they have an idea of what processes are poorly represented. They can have an informed opinion on what that means for projections.
and, oh yes, if you want the total sea level rise, you have to include expansion of the oceans due to warming (22 cm), glacial melt (15 cm) and water mining as in pulling water out of deep reservoirs (5 cm).  Add it all up as Grinsted does and one is well over a meter for worst case and damn close for best guess.


The Old Man is back said...

I was going to comment, but had so much to say I decided instead to post a response to this, trying to put the significance of SLR into context. I don't think most people 'get' why it matters.

EFS_Junior said...

So, the more uncertain we are about the future, the more we should prepare for that more uncertain future?

Sounds sort of tautological, so in keeping with today's theme ...

Why stop at a 90% uncertainty interval?

Let's all go with a 100% uncertainty interval.

Infinite SLR, by 2014 even!

Also, last sentence, cleaver wording, or not, we decide?

worst = maximum (p = 5%)
best = mean/median (p = 50%)
least = minimum (p = 95%)

The minimum SLR (as opposed to worst/maximum) is still much less than one meter, even using Grinsted's "dartboard" ice sheet numbers.

One commonly refers to water levels in terms of probability of exceedence, so Grinsted's 95% is really 5% probability of exceedence.


Tenney Naumer said...

Such a pity Yulsman could not have read Fergus' latest post before he wrote that gobsmacking fluff the other day.

Anonymous said...

Hi Eli,

In the unlikely event that I recall correctly, AR4 and earlier reports assumed no sea rise at all due to glacial melt. This apparently because of uncertainty.

The IntergovernmentalPCC produces ludicrously conservative predictions. Lest we forget, that was always its raison d'etre.

a_ray_in_dilbert_space said...

Scientific consensus almost always produces conservative opinions, whether the subject is nukes (linear-no-threshold), climate or where to have dinner.

That is one reason why the denialists rejection of consensus is so ludicrous.

EliRabett said...

6/11/13 1:05 AM: (please take a number:).

Not precisely, they said that their estimates could and did not include ice cap (?) melt because there was insufficient published basis for evaluating it. Read closely, that meant that their estimate was a lower bound.

No one reads closely.

Susan Anderson said...

What gobsmacking fluff? This?

The Old Man is back said...

Thank you Tenney :*