Interesting news in that a group of GW Deniers, including Canadian Ross McKitrick, have managed to place a paper, entitled Does a global temperature exist? in a real, peer-reviewed journal (the Journal of Non-Equilibrium Thermodynamics), although from what I can gather it is a relatively "low rent" journal where you send the papers other journals won't take.
From their news release at Science Daily:
It is generally assumed that the atmosphere and the oceans have grown warmer during the recent 50 years. The reason for this point of view is an upward trend in the curve of measurements of the so-called 'global temperature'. This is the temperature obtained by collecting measurements of air temperatures at a large number of measuring stations around the Globe, weighing them according to the area they represent, and then calculating the yearly average according to the usual method of adding all values and dividing by the number of points.
The authors claim that:
"It is impossible to talk about a single temperature for something as complicated as the climate of Earth", Bjarne Andresen says, an expert of thermodynamics. "A temperature can be defined only for a homogeneous system. Furthermore, the climate is not governed by a single temperature. Rather, differences of temperatures drive the processes and create the storms, sea currents, thunder, etc. which make up the climate".
He explains that while it is possible to treat temperature statistically locally, it is meaningless to talk about a a global temperature for Earth. The Globe consists of a huge number of components which one cannot just add up and average. That would correspond to calculating the average phone number in the phone book. That is meaningless. Or talking about economics, it does make sense to compare the currency exchange rate of two countries, whereas there is no point in talking about an average 'global exchange rate'
Well, maybe, but the real problem with this claim is that it proves too much, can in fact be employed to prove that the concept of an average anything is meaningless.
Consider housing values (which I will use because statistics are easy to get hold of). For example, in January 2007 the average price of a resale home in Canada came in at $282,844. But wait! You can argue that it is meaningless to talk of an average house price for Canada, because Canada "consists of a huge number of components which one cannot just add up and average". After all, people don't purchase a house in Canada, they purchase a house in some specific locale, a Metropolitan area like Calgary or Edmonton.
So lets consider Edmonton, where the January average was $321,307. But wait! That figure, one can argue, is also meaningless, because people don't really buy in Edmonton. Rather, they buy in one of the "huge number" of sub-markets that make up Edmonton. But wait! They don't really buy in a sub-market either, they buy on a particular street or in a particular building. And so and so forth!
So we have managed in a few words to rubbish the concept of an average house price.
Now, in their paper the authors attempt to short-circuit this line of criticism by suggesting that we only employ the concept of averaging where "it makes sense". But where averaging "makes sense" is highly context sensitive. An example mentioned in the paper is the "average height" of a population. Well, if you are dealing with the global tree population, which includes both maples and pines, and both adults and saplings, you run into the same problem. You also run into the same problem if there is a wide enough distribution of individual heights across adult trees of the same species. (Note that this is not an entirely hypothetical example. There are occasional arguments in paleontological circles about the utility of average size statistics: is the average size of a Tyrannosaur a useful conceptual tool?).
So once again, accept the paper's logic and the whole concept of an average value turns out to be mythical.
Way to go, fellas!
A pre-print of this paper can be found here. A much more intensive critique can be found here, including some nice comments on an earlier version of the argument I am making in this post.