We have precise data for Arctic sea ice extent from satellite observations over about the last 30+ years, and a pretty good estimate for the last century or more from ship and aircraft observations. The data show clearly that Arctic sea ice has declined dramatically over the last few decades, a decline the like of which has certainly not been seen for over 100 years prior to that. In addition, a recent survey of available proxy information — not a reconstruction, but a thorough review of the evidence — tells us that the modern decline in Arctic sea ice is “unmatched over at least the last few thousand years and unexplainable by any of the known natural variabilities.”
Now there’s even more evidence of the shocking state of the Arctic sea ice. New research reports an actual reconstruction of the extent of Arctic sea ice over the last 1,450 years (Kinnard et al. 2011, Nature, 479, 509-512, doi:10.1038/nature10581).
Kinnard and his coworkers collected 69 proxy data records from the Arctic region, mostly from ice cores (using oxygen isotope ratios, percentage of infiltration ice, and sea-salt ions) but also including tree rings, lake sediments, and historical data of sea ice observations. Many of the proxies have information about physical properties other than sea ice extent — especially temperature — but the analysis method chosen (partial least squares) enabled the authors to identify modes of variability in the proxy data which are distinct from the temperature signal, and which correlate with observed sea ice extent. This method also helps overcome the “collinearity problem,” in which multiple proxies are so similar to each other that they provide overlapping information.
The regression correlated proxy data with late summer (August) sea ice extent, both from hemispheric data and from additional data for the Russian Arctic. Kinnard et al. found that their reconstruction had skill from the year 561 onward, i.e., over the last 1,450 years. The “hockey-stick” shape of the history of Arctic sea ice is remarkable (smoothed with a 40-year lowpass filter):
Uncertainty levels are greater the farther back in time one goes, because there are fewer proxy records to use for reconstruction. Nonetheless, the remarkable decline in the late 20th century is far beyond anything seen before it, both in magnitude and duration.
Prior to the recent decline, there were periods of sustained greater sea ice (about 1250 to 1450, and 1800 to 1920) and periods of sustained lesser sea ice (before about 1200). The minimum of sea ice before the industrial revolution was even earlier, around the year 640. There were also two later episodes of sea ice decline, in the late 16th and early 17th centuries, but none of these other declines even comes close to the “falling off a cliff” of sea ice we’re seeing today.
Surprisingly, sea ice shows signs of decline during the “little ice age” when it might naively be expected to increase. The authors suggest that transport of heat into the high Arctic may have been responsible, so this episode would represent one version of “heat piracy” in which the cooling in Europe and elsewhere was accompanied by warming of the Arctic. Likewise, the pre-industrial minimum in the year 640 coincides with what they refer to as the “dark ages cold period,” and may similarly represent the diversion of heat (perhaps by ocean currents) to the Arctic.
But the modern crash of sea ice extent corresponds to a warm period, both hemispherically and globally, so it’s not a case of heat migrating from one region to another, but of — how shall we say? — global warming. The authors conclude:
In the present state of knowledge, anthropogenically forced (‘greenhouse gases’) warming stands out as a very plausible cause of the record atmospheric and oceanic warmth of the recent decades, which may soon lead to an ice-free Arctic Ocean in summer.
Very nice overview, Tamino. I’m very curious to see what the next two melting seasons bring.
I’m not. A couple more seasons of “See! I told you so!” is like cutting off your nose to spite your face.
Thanks, Tamino. As I think you are aware, it’s been common for some to claim–with no evidentiary support whatever, barring the odd citation of Gavin Menzies or some such–that the Arctic sea ice has been just as low as now many times previously. This will be convenient to brandish during future iterations.
[Response: Making a claim with no evidence to support it — that’s how they roll.]
Jebus, I look at that and think we’re so screwed because, by the time we get our act together, it’s going to be too late.
But … but … those photos of the USS Skate! What other evidence do they need?
A real photo of USS Skate after she just surfaced at the Pole March 17th 1959, two days before polar dawn, is always worth a bookmark.
Damn you! Cherry picking photos that show the real state at the pole!
You’ll never be a WFUWT hero !!!
I do like that!
Enquiring minds would suspect Photoshop to be involved…
After all, it must be a picture taken during the Arctic summer ice recovery of 2011 (because we know SG is NEVER wrong) but tinted to make it look older. Those same enquiring minds know that there was still no ice at the pole in March 1959 (it had all melted in the 30’s and 40’s when the Vikings were plantation farming in Greenland).
The recovery continues (and there is no man behind that curtain)…
Being a “skeptic” is fun & easy!
Gavin Menzies book 1421, Chinese Arctic exploration in the 15th century. Some of the many fantasies in the book include, “The Chinese established colonies in many countries including Australia, New Zealand, California and Mexico. They introduced horses to the Americas, rice to California, chickens to South America, coffee to Puerto Rico and sloths to Australia and mined for lead. Sea otters to New Zealand, and maize to the Philippines.” The result of misreading just about everything he read .
Even then Menzies only writes of the sub Arctic, Chinese whispers?
Can you post the Antarctic sea ice trend as well for comparison?
@ Brian Rookard
Perhaps you could look at a globe and note the latitudes of the areas covered by floating sea ice in the Arctic vs those covered by floating sea ice in the Antarctic.
That should be all the comparison any true skeptic should need.
The CT Antarctic SIE anomaly curve can be inspected here:
Better would be the model predictions made 30 years ago for the Antarctic compared with the Antarctic sea ice trend.
Because, you know, otherwise you’ll say “increased sea ice proves climate science is a fraud!!!!” even though warming was predicted to increase Antarctic Sea Ice for several decades …. a prediction made long, long ago.
Only in the denialist mind can a prediction of science that comes true be construed as proving the underlying science false …
Our host may think your question is innocent, and not approve my response … but my mind’s made up.
Really? I thought that the early climate models had made incorrect predictions for the Antarctic, but that we were fairly confident that this was because they didn’t include the Stratosphere properly, and so couldn’t model the effect on the Antarctic circulation created by the ozone hole.
About half of the models for CMIP5 will have a well-resolved Stratosphere, so we should see a clear difference between the two sets of models in their Antarctic predictions – though if we assume ozone recovery over the 21st century then their predictions will tend to converge in any case.
Arrghhh. That’s a graph to make the bowels loosen, no doubt.
Over at Climate Audit S.M. said, “…the reconstruction is a linear combination of proxies…” and claims (for instance) that the long ice core O18 series can’t produce the ‘hockey stick’ like graph in Kinnard 2011.
I’m not a statistician, but my understanding is that EOF analysis looks for relationships between data sets. These relationships may or may not be linear. The power of EOF is that it can find relationships that are not intuitive or obvious.
The Kinnard reconstruction is for sea ice extent. Since most of the proxies used in Kinnard are temperature proxies and all the non-documentary proxies are land-based (ice cores, tree rings, and lake varves), the reconstruction has to develop a relationship between sea ice extent and land temperatures.
Perhaps I’m ignorant, but I find it unlikely that a linear relationship exists to describe a land-based temperature x and the simultaneous sea ice extent y. I find it more likely that a complex non-linear relationship exists and that this is what Kinnard et al have found and described.
Is S.M. correct that the reconstruction is a linear combination of proxies, or is there far more to it than that?
[Response: The reconstruction is a linear combination of proxies. But it can produce a hockey stick. In fact it did.]
Sorry, from ignorance I poorly phrased my question. If the reconstruction is based on a linear analysis of the data, how are possible (likely?) non-linear relationships captured? Or is there an a priori assumption that the relationship is linear?
I’m not doubting the ‘hockey stick’ nature of the decline in Arctic sea ice extent. That’s apparent from 150 years of observational data. In fact, observational data would indicate the decline is occurring even faster than the reconstruction would have us believe. The greatest reduction in sea ice extent has occurred during the last 10 years – years not covered by the reconstruction. The vertical scale would have to be *doubled* to cover the last 10 years of observational data.