Friday, December 18, 2009

The Data on Global Warming

Last week a regular reader raised some questions about the significance of the data presented in this chart:
The reader's comments:
I find that chart not very meaningful. First off, the Y-axis has a span of only a few degrees F, so small changes are made to appear artificially large. Second, and more important, reliable global temperature data prior to the 20th century is just about non-existent. That is why proxies, such as ice cores and tree rings are used.
Let's take a look at these two questions. The range of the chart is narrow, but is not artificially so; it encompasses all of the data. The significance of the data is found in the recent departure of temperatures form the narrow range that prevailed for 1,300 years. In contrast to the the cumulative weight of these data, the skeptics who repeat the canard that the earth is cooling rely on data from one anomalous year to make their point. Clearly something very different has been happening in recent years, and it can't be attributed to sunspots (another explanation skeptics offer), which follow an eleven year cycle.
As for the techniques of using ice cores and tree rings, scientists use direct measurements to calibrate the indirect data. Overall, we are looking at many thousands of measurements, collected by teams of researchers all over the world. Again, this is why science relies on peer review of data and conclusions.


Anonymous Edmund Dohnert said...

Tom -

You are of course referring to my previous comment on the subject.

Though I think global warming is real and that we face a real potential problem, the graph in question and the data it represents are in my view still highly dubious.

Regardless of how much data is represented by the graph, the range of the vertical axis is still very narrow, being only 1 degree F in the positive direction and 1.5 degrees F in the negative direction. For the sake of argument, let us say that the average global temperature at any given time is on the order of 50 degrees F. As such, the amount of graph in the positive direction (1 degree) is only 2 percent of the average temperature. Thus, small differences, some statistically significant and some not, are highly exaggerated.

More important, and far more problematic, this graph mixes (and one might say, confuses) several very different types of data sets. From about 1950 onward, the data on global temperature is quite good. From 1900 to 1950, it is a bit spotty, but still relatively decent. In the 19th century there was a lot of temperature data, but it was mostly confined to the US, Europe, or some parts of the world colonized by Europe. Prior to that, it is almost entirely inferred from surrogate or proxy measurements, such as ice cores and the like.

Ice cores can tell you a lot about what is going on in the arctic regions, and tree rings can tell you some things that are going on in the temperate zone. But that leaves out about 70% of the earth's surface area, i.e., the oceans. Ice cores can tell you what the CO2 concentration was and how much snow fell in Greenland in the 1300s, but it can't tell you what the average temperature was in the Sahara or the Australian outback or most of the rest of the world for that matter.

Tree rings can tell you whether the year 1326 had a better growing season that the year 1327, but it really can't tell you what the average temperature was. That is inferred using a variety of assumptions and models, but it is not (and cannot) be based on accurate measurements.

So, what the graph does is combine several centuries worth of inferred global temperature values of an highly inaccurate nature, with less than a century's worth of reasonable accurate data. Essentially, up to about 1900, the graph is plotting apples, but then it starts plotting oranges, and then finally around 1950 it starts plotting cumquats.

I conclude that this graph is more of an advocacy tool than an impartial scientific representation.

6:13 PM, December 18, 2009  

Post a Comment

<< Home