I'm meltinnnnng ....

Some commenters seem puzzled by my conclusion that a couple of recent studies of melting north polar ice could mean an ice-free Arctic within 13 years. I will agree that it does seem rather extreme, but the data support such a conclusion, as a responsible estimate of the near-term end of range of time values. Perhaps part of the problem can be traced to a rather poorly designed graph, drawn from data in a paper (subscription only) published Tuesday in Geophysical Research Letters. Allow me to explain:

The paper, by Julienne Stroeve, Marika M. Holland, Walt Meier, Ted Scambos, and Mark Serreze of the National Snow and Ice Data Center at the University of Colorado, basically says that previous computer models are too conservative, and that actual observations suggest the Arctic is melting much faster than is widely thought. Their paper contains these conclusions:

Observations indicate a downward trend in September Arctic sea ice extent from 1953-2006 that is larger than any of the IPCC AR4 [Intergovernmental Panel on Climate Change Fourth Assessment] simulations, and current summer minima are approximately 30 years ahead of the ensemble mean model forecast.

The IPCC AR4 models indicate with the "business as usual" ... may be realized anywhere from 2050 to well beyond 2100.

So is it fair to simply subtract 30 years from the IPCC range and you get 2020 to 2070? On first glance, no, as 30 years ahead of the mean is really just a smidgen ahead of the some of the more alarming computer models run in the past. They see summer ice disappearing in 2050. But you can get there by looking at the bigger picture and the study's smaller details.

Scientists used to suspect the ice was melting at 2.5 % per decade. But the new data, using much more reliable observations, produced a rate of more than three times that -- 7.8 % per decade (1953-2006). So whatever algorithms the computer models are using need to be adjusted to take into account this accelerated melting rate.

Two years ago, it was generally accepted that the pole would be free of summer ice by about 2080, give or take. Last year that was moved up to 2048, give or take. Now we know that those predictions, based as they were on much slower past melt rates, are far too conservative. If the seasonal ice is melting three times faster, then it is not unreasonable to posit that the time between now and and ice-free summer is a third of what was once though. We used to think we had 40 years. We might only have 13. Even if we look at the top end of the models, we're only talking 33 years from now.

Remember that current observations are already 30 years ahead of the ensemble mean, and the faster melt rate means we're going to continue to get farther and farther ahead of that mean.

As I wrote in the original post, this agrees with observations from Cambrige University's Peter Wadhams, who just returned from a submarine expedition to the region. According to media reports, "Wadhams' measurements indicate that the thinning was already approaching 50 percent and that the ice could disappear by 2020."

Next, there's the fact that, according to the authors, the rate of sea ice loss over the last three decades was considerably greater (-9.1%/decade) than the rate of the previous decades (-7.8%/decade). In other words, the loss rate appears to be growing. As they write: "Either way, it appears that impacts of GHG loading on Arctic sea ice in September are strong, and growing..." This means that you can't just extrapolate linearly from past observations. The slope will more likely increase (steepen).

This makes sense because so many glaciologists and climatologists expect the reduction in the Earth's albedo that will accompany a shrinking polar ice cap to produce a positive feedback loop that effectively accelerates future melting.

Unfortunately, this graph (click to enlarge), which accompanies a press release from the National Snow and Ice Data Center, and its oddly chosen perspective, doesn't really help.

It actually makes extrapolating the future trajectory of the observations more difficult than necessary. So, I called one of the authors, Julienne Stroeve, and asked for permission to reproduce Fig. 1 from their paper, which is what the above graph is based on. (After what happened with Shelley Batts at Retrospectacle, I'm taking no chances.) The journal isn't allowing bloggers to post figures from the paper, but she did agree to give me similar graphs she generated from the same data. Here is one. Again, click to enlarge.

The red line represents actual observations. The heavy black line is the mean of a variety of computer model predictions, flanked all the various model's predictions.

It is important to remember that sea ice cover trends historically exhibit "strong imprints of natural variability." So even with such convincing data, a good portion of what's been seen is likely normal. But because we really don't know how those normal cycles tend to go, the future is still largely uncertain.

Nevertheless, it is only prudent to operate on the assumption that something is amiss up there in the hyperborean latitudes. If we do extrapolate the red line into the future, taking into account some element of continued acceleration of the melt rate, and factor in an appropriately expanded cone of uncertainty, you get this (my modification):

All of which explains why I am comfortable with the notion that this paper supports the possibility that all the summer ice will be gone as as soon as 2020. Hopefully things aren't that bad, but it's based on the available science and it's not an unreasonable guess.