GRACE puzzle

The paper-for-today is Isabella Velicogna and John Wahr, Measurements of Time-Variable Gravity Show Mass Loss in Antarctica. Both Chris Mooney and Kevin Vrames have things to say about this washingtonpost write-up; I'll try to talk about some different aspects.

First of all, this is only 3 years of data; to make any sense of it, you have to assume that this is representative of the long term trend. Suppose we do this. What washpost, and the Grauniad, and every one else seems to have missed is that this just tells you, if we take the paper at face value, that Antarctica is losing mass, at 0.4 mm/y of sea-level-equivalent. I can't remember what the TAR thought; this suggests that they were unable to decide a good value, though this fig suggests they settled on a net negative contribution. But change from -0.2 mm/y to +0.4 mm/y doesn't mean the true value has changed, just a new estimate of what the value might be; if you read the newspapers you get the impression that Antarctica has suddenly started melting.

Another point (see last IPCC link) is that its quite hard to match the *observed* sea level change to the various components; switching Ant from -ve to +ve would help resolve this.

How does this affect predictions that Antarctica will become a net relative sink in the future, due to extra precipitation outweighting increases in melt? Not at all, I should say. That still remains valid.

Lastly, how much should we trust these results? 3 years of data is not much; longer trends are needed. Also, James Annan has some wise words about trusting too much on the latest paper; these things really need time to settle. But all that aside, note that the results depend very heavily on the adjustment for isostatic rebound. This is because Antarctica is smaller now than at the height of the last glacial; hence lighter; hence the rock underneath is slowly moving upwards (post-glacial-rebound; PGR). GRACE measures gravity anomalies; ie weight of rock and snow (errm, and actually also weight of nearby water masses, which could be a problem on so short a scale as 3 years). So if the rock is going up, that needs to be subtracted to get mass of ice. The "problem" is that this is not a small term; see the papers figure 2, which shows that the trend is flat, without the correction for the rebound term. As the authors say:

The main disadvantage of GRACE is that it is more sensitive than other techniques to PGR; in fact, our error estimates are dominated by PGR uncertainties. As more GRACE data become available, it will become feasible to search for longterm changes in the rate of mass loss. A change in the rate would not be contaminated by PGR errors, since the PGR rates will remain constant over the satellite's lifetime.

This doesn't make the results invalid, of course; its just that an accurate assessment of PGR is hard, too.

[Update: the Scotsman Antarctic ice sheet is melting at rate of 36 cubic miles a year, says study sez much the same as the Grauniad (cos they got the same press release...) but does quote "our" David Vaughan with However, Professor David Vaughan, of the British Antarctic Survey, said the figures produced were not radically different from previous estimates, despite being produced using a totally different method. Normally scientists base their calculations on measurements of the height of the ice. And he added: "The record [showing the loss of ice] that's been published is over a very short period. We'd really like several decades of record to be confident the changes we're seeing are long-term trends."

Oh, and I should have mentioned my thanks to John Fleck]

[Another update: I should have realised that the TAR fig has Gr/Ant (recent) as well as a *postive* contribution from Gr/ANt (long term). Which is mixed up with the PCR term. So this may not help as much as I suggested :-(]