Friday, March 11, 2011

Latest Ice Sheet Mass Balance Stats (Take 2)

Note: I wrote a short post on this subject yesterday, but it contained a significant error.  This is a longer exploration of the issues that is hopefully correct.

There is a new paper out: Acceleration of the contribution of the Greenland and Antarctic ice sheets to sea level rise by Rignot, Velicogna, van den Broeke, Monaghan, and Lenaerts.  (Aside: if you aren't associated with a university or major research institution, you will likely have to pay \$25 to read it.  I find it intensely irritating as a taxpayer that I have to pay \$25 to read a five page paper that describes publicly funded research, but there it is).  The paper is likely the new state of the art in estimates of how much ice is being lost from the big ice sheets on Greenland and Antarctica.

The important background to the paper is to realize that there have been two completely independent approaches to measuring the total loss/gain of the ice sheets.  The first of these was more of a bottom-up approach that involved estimating the amount of snowfall on the ice sheet from weather models, the loss due to surface ablation (ice vaporizing etc), the speed that glaciers were moving at (from satellite pictures), the height of the glaciers, etc, and performing a detailed cell-by-cell estimate of how much the mass in the ice sheet was changing, which then needed to be integrated across the entire ice-sheet.  Eric Rignot has been a key name in this approach.

The second approach has been to look at measurements of the earth's gravity field measured from NASA's Grace satellite and performing mathematical analyses of the small shifts in it that occur over time to isolate the changes that are due to changes in the big ice sheets (you can imagine you'd need to filter out all manner of other things - daily tides, changes in other glaciers and snow fields, changes in sea level due to atmospheric pressure systems, changes in ocean currents, etc).  Isabella Velicogna has been an important name in this approach.

Both of these are obviously enormously complex calculations that require years of work, and they didn't come up with exactly the same answers.  In this paper, scientists who have worked on both approaches got together and reconciled them.  They made various tweaks and corrections by looking at the differences.  To give one example, the mass balance people had been assuming that the grounding line where an outlet glacier that drained the edge of the ice sheet left the sea floor and started to float was in a fixed position.  This isn't really true - as the glaciers speed up and warm water gets under them, the grounding lines are moving inland, and this caused an important discrepancy with the gravitational measurements in some places.  So they applied some corrections.

Here is the comparison of the two results, separately for Greenland (top) and Antarctica (bottom):

The blue/black is the mass balance estimates, and the red is the gravity estimates.  In both cases what is being measured on the y-axis is the rate of change of mass (ie how many gigatonnes of mass the ice sheets lost in a particular unit of time).  The data are monthly, but expressed at annual rates (gigatonnes/yr).  You can see that the gravity measurements and the mass balance approaches are now in broad agreement - not perfect, there are clearly still systematic differences, but those differences are now much smaller than the phenomenon that both are seeking to measure.

So, presumably, now that these two approaches have been reconciled, we should have higher confidence in the results than in previous papers.  And at least the mass balance approach data now go back to 1992.

The picture that emerges is, to my mind, somewhat puzzling.  In particular, if you look at the black trend lines, they cross the x-axis (zero) in Greenland in about the end of 1993, and in the case of Antarctica in about 1991 or so (if you imagine continuing the line on the lower plot back past the left hand end of the graph).  Zero here means "no change in mass balance".  So the implication seems to be that prior to the 1990s, the ice sheets were not losing mass.  Then, both ice sheets crossed some kind of non-linear threshold at almost exactly the same time and started to lose mass at an accelerating rate.  In particular, the rate of mass loss has been increasing roughly linearly (there's some noise, but no sign of a systematic departure from that straight line in both the graphs above).

Since the rate of mass loss is linear, the total mass lost (and thus sea level raised) is quadratic.  It's not obvious to me why ice sheet loss would be expected to turn on suddenly and then be quadratic, and so when we extrapolate this trend (which I'm about to do), that should be taken with a large grain of salt.  In general, one of the lessons of this whole episode is that non-linearities matter a lot in the climate, and the system can and perhaps will continue to throw up surprises that climate scientists don't anticipate.

In particular, given that the climate has been warming slowly and unevenly for the last 150 years, and given that Antarctica and Greenland are exposed to completely different parts of the ocean/atmosphere system, why would both cross some kind of threshold at almost exactly the same time?

Anyway, here is the combined trend for both ice sheets:

To compare this to other work on sea level, I took the paper's quoted sea level rise contribution in 2006 (1.3mm), and scaled the line above (with trend of 36.3 Gt/yr2) by that, and then extrapolated that out to 2100.  Then I took what I take to be the state of the art estimate of future total sea level rise, which comes from Vermeer and Rahmstorf, 2009.  That paper builds on an earlier one by Stefan Rahmstorf which simply proposed looking at the rate of global sea level rise as being driven by the difference between global mean temperature and pre-industrial temperatures.  It turns out that this kind of approach can do a very good job of explaining the existing sea-level data.  Anyway, comparing these projections with the latest data from the Rignot et al paper, we get the following:

Here, the y-axis is total sea level rise (in mm), and the broad colored bands are the Vermeer and Rahmstorf estimates under various emissions scenarios.  The yellow band is under A1F1 A1FI which is an all-out fossil fuel intensive scenario, and the green is B1, which involves continued economic growth, but people turning increasingly to conservation and renewable energy over the course of the twentieth century.

The small vertical colored bars to the right hand side (at about the 400mm line) represent the sea level estimates from the IPCC AR4 report for 2100.  Those estimates explicitly excluded ice sheet dynamics, and thus are widely considered to be underestimates.

The blue line is the new estimates from Rignot et al, which only includes contributions from the Greenland and Antarctic ice sheets.  In addition, sea level rise will also be driven by thermal expansion of the ocean, and melting of mountain glaciers and ice-fields.  Those additional processes are effectively included in the Vermeer and Rahmstorf estimates (colored bands).

So, at risk of oversimplifying a bit*, you can think of the blue line as being the thing excluded from the IPCC 2007 estimates.  And you can imagine adding the level of the IPCC estimates (thin vertical bars) to the ice sheet estimate (blue line) and being in the ballpark of the Vermeer and Rahmstorf total estimates (colored bands).

So, it's in this sense that I say the science is settling down a bit from the discovery that ice sheet glaciers started to speed up much more than anticipated in the 1990s due to global warming.  It looks like we are heading for ballpark a foot of sea level rise by mid century (pretty much regardless of what we do) and 3-5 ft by the end of the century (depending on emissions and uncertainties in the climate sensitivity).

That will be catastrophic for some low lying places (certain Pacific islands, Bangladesh).  It will increase the risk and severity of events like Hurricane Katrina or the Japanese tsunami today.  When all the world's coastal cities have significant portions of them below sea level behind levees, the vulnerability to those kinds of natural disasters will be greater.

Still, if you think of 9 billion people in 2100, to a first rough approximation, as 9000 urban districts of 1m people each, it's clear that the overall juggernaut that is civilization is not going to be stopped by the loss of a few tens of those to natural disasters.  It's in this sense that I say sea level rise currently doesn't look like a serious risk to global civilization.  It will nibble at the edges here and there, but it doesn't presently appear to have any potential to collapse the entire thing.

The one big caveat is this: the climate system is non-linear and keeps surprising the climate scientists. There may be more surprises ahead.

* In particular, the IPCC estimates were intended to include surface mass balance estimates on the ice sheets, which are also included in the Rignot et al paper.

Greg said...

Quibble: that's A1FI, A1-Fossil Intensive. (I noticed Joe Romm making the same mistake - certain sans-serif typefaces are to blame, I guess.)

I agree that changes in sea level don't constitute a global threat. I'm much more concerned about changes in ocean chemistry. That could have severely non-linear effects. The possibility of a "flip" to complete thermal stratification is also a worry.

glacierchange said...

You have but to look back to the papers in the 1990's on the topic to realize, we were struggling to assess the mass balance of either ice sheet. Based on data available before 1990 ie. Vaughan et al., (1999). The key is the slope of the volume loss trend line, which we cannot know going further into the past as the remote sensing data is too scant. Field data was also scant. I published a paper on Jakobshavn in 1989 that indicates its equilibrium state at that point. At that point we were studying Jakobshavn because we saw it as an analog for Pine Island Glacier in West Antarctica.

Stuart Staniford said...

Greg - whoops, thanks, fixed.

Greg said...

What we are not including in our estimate are new positive feedbacks. And while we can guess, we don't know their impact.

When the Arctic is ice free, what will be the impact on the permafrost land nearby?

What if 10% of boreal forests burned down one year?

What if peat ignited in Alaska or Siberia and burned continuously for decades?

What if the clathrate gun theory proves true?

Watching trends and extrapolating is fine but what will doom the the biosphere is probably not a member of a smooth curve, even quadratic.

Civilization, more tender than the biosphere, doesn't have to be brought down by few bad storms, just enough fear to international insurers to stop covering international trade or new developments.

Insurance is the oil in the engine. Uncertainty and risk are the now quickly depressing brake on economies.

Bruce said...

Hi,
Thanks so much for such informative articles. I've really enjoyed (maybe not the right word ... valued?) reading them. However it occurs to me here that you might have underestimated the danger of the sea level rise: most towns are built where rivers meet the sea. If we put up 1.5m barriers to stop the sea encroaching, we may find our towns flooded by fresh water rather than sea water, but the effect would be much the same.
Obviously we could extend the flood barriers a few km up every river, but that would a) be an awful lot of work and b) doubtless have further knock-on effects.