Over on Climate Audit, Steve McIntyre did a post about what HadCRU recently did to announce an “error” in their temperature series due to the significantly colder temperatures worldwide in January and February 2008. They made a bold front page announcement about it which you can see here and at left.
We have recently corrected an error in the way that the smoothed time series of data were calculated. Data for 2008 were being used in the smoothing process as if they represented an accurate estimate of the year as a whole. This is not the case and owing to the unusually cool global average temperature in January 2008, the error made it look as though smoothed global average temperatures had dropped markedly in recent years, which is misleading.
For their influential graphic showing smoothed temperature series, they used a 21-point binomial filter (this is reported) extrapolating the latest number for 10 years. This obviously places a lot of leverage on January and February temperatures. As has been widely reported, January and February 2008 temperatures are noticeably lower than last years.
He also made a graph to show the differences before and after HadCRU made the adjustments to their data:
Michael Ronayne pointed out to me today that this is not the first time HadCRU has modified their online graphs. In happened before in 2000 as the late John Daly reported:
CRU found that even with their disputed surface record, there was a sharp cooling in both hemispheres from a peak in 1998, but their global graph did not reflect this – instead it shows a resumed warming.
Back at Climate Audit, in comments, somebody asked some pointed questions about why this happens, suggesting less than honorable motives. Steve McIntyre doesn’t think so and writes:
The point is that these institutions seem far more alert to errors causing something to go down than to errors that cause something to go up.
I agree, and would attribute it to “expectation bias” on the part of the HadCRU data gatekeepers. Since they are English, I’ll use the tea analogy.
You are making tea. You put water to boil on the stove, light the fire, and set the teakettle on the burner, see that all is well, and go about your business.
You look over from your desk, you see the burner going, the kettle is making the pops and creaks as the metal expands due to increasing temperature. All is well, the temperature is rising.
In two minutes, and you begin to hear the chorus of small bubbles forming on the bottom. No need to look over, all is well. The temperature is rising.
In another minute, you hear bubbles, no need to look to see thin wisps of steam rising from the spout, all is well. The temperature is rising, water should be ready soon.
30 seconds later, the whistle begins, and you know the heating process (AGW) went perfectly. The water temperature went up as expected and there was no need to check the kettle or the stove during the process because the end result was expected based on the starting set of conditions.
But if the burner had gone out, just before the whistle, you wouldn’t notice it, for some time, until you realize the whistle never came. Then you’d get up from your chair to do something about it. Ah, the burner went out, the water is cold, we’ll move it to another burner that isn’t faulty.
All is well.
Expectation bias in temperature rise, the Lipton Tea of climate science.