Antarctic Trends, Graphics and Discussion
Posted by Jeff Id on December 9, 2010
On Climate Audit, in response to a request by Hu McCulloch, Ryan posted the following improved graphic.
WordPress likes to help out with resizing images, so for the really big one go here.
Compare that to the Steig et al. reconstruction below which was used as the cover for Nature.
Back in the day when this work was still fresh in blogland, we made the point repeatedly that the warming in the peninsula was being spread around the continent. On searching for a better jpg of the original Nature cover I ran across this reminder from Gavin Schmidt.
James Martin says:
I read today a claim that in the paper published recently by Dr Steig et al. in Nature regarding the Antarctic warming trend, there is a weighting problem. They claim that most of the weighting comes from the peninsula stations, which represents a relatively small part of the continent.
I was wondering if this is in fact the case? It doesn’t seem likely, but could you comment on this at all? If these assertions are left unchecked, before you know it they’ll be taken as fact.
[Response: The point of the Steig et al paper was to use spatial correlations in recent data to look at how under-sampled parts of the continent likely changed over longer time periods. Those correlations will necessarily weight different stations differently as based on the physical characteristics. The analysis you saw is simply a fishing expedition, an analysis of what the calculation is doing (fair enough), combined with an insinuation that the answer is somehow abnormal or suspicious (not ok). But how is this to be judged? What would be normal? No-one there can say and they would prefer simply to let people jump to conclusions. It’s kinda of typical of their tactics, but not a serious scientific point. – gavin]
My bold of course, while it is funny looking back at the usual RC commentary, there is a point yet to be made here. These methods will always produce that kind of bleed of trend in the result. The trick is to minimize it. The spreading happens because they work by infilling missing data based on correlation of surface and satellite datasets, both of which contain noise. This has the advantage of distributing information according to real world weather patterns rather than according to spatial location with the disadvantage of a bit of information spreading. There were a number of mathematical improvements made in our recent paper which Ryan will explain in time, but one of the main ones was that our method used surface station trends only and spread them across the continent using the – trend noisy but spatially reasonable satellite surface skin temperature data. Yet no matter how improved the method is, it will always have some of this problem.
My own contribution to this work was fairly limited. Some commentary on methods, some minor editing and the original version contained some area weighted reconstructions similar to those presented here and below. It was later removed from the SI after the first two reviews required us to completely change methods well away from Steig’s. You can see that the pattern of trends is reasonably well matched to the published version above even though the image below used a different set of surface stations. The difference though that catches my eye is the containment of the peninsula (image below) just to the finger region whereas there is a bit of trend bleeding into the mainland created by the more sophisticated methods.
A visual comparison of the above with our final results below can give a better idea of what the blending causes. The peninsula region warming spreads slightly past the thumb into the mainland. However it is far more defined than Steig 09 and much closer to the actual surface station trends shown in the closest station figure immediately above.
This next image was posted here also in the August timeframe and while the method is similar to what was published, it isn’t the same. The image from the published paper above does a far better job containing the peninsular trends than the one below. However, despite the fact that different stations and methods were used, you can see that they are all quite similar, except for Steig et al.
In all I would guess that hundreds of tweaked methods were used during the learning process. What is learned when so many methods are tried is that the result is robust (ouch) to the method used and the basic result is something you can be confident represents the actual data. On a final note, it was amusing in the review process where we were pushed to make conclusions in comparison to climate models and to support the conclusions of Steig 09 as conditions of publication. I’ve read a few papers on the models of the Antarctic but know very little on the topic. I was and still am very much confused as to why the ‘models’ mattered at all when solving a math problem and found it rather interesting that climatologists would ask engineers, mathematicians and physicists to interpret a result in terms of weather patterns of a climate model.