Mar 31 2010
Ross McKitrick has a long narrative up on how he has attempted to correct the sloppy math and unproven assumptions made by the IPCC and CRU (H/T Bishop Hill). Â As you read this you should also be aware that inside the CRU data dump from last fall was a smoking gun document which clearly indicates CRU knows its data is incapable of detecting AGW levels to a tenth of a degree. To do so requires temperature data to be at least that accurate or better. You can’t measure to the inch with an unmarked yard stick.
I mentioned the document months ago in this post (and others) as evidence that the CRU temperature data DOES NOT have the accuracy required to support the IPCC claim to detect tenth of a degree changes over decades over the entire planet. The document is very interesting in that it shows for one year in the past (1969) the uncertainty in the CRU temp data grids (click to enlarge).
The data clearly shows that for 1969 the CRU temperature data sampling errors ranged from 1Â°C for most of the world and up to 4.5Â°C for a lot of North America and Russia.
As the McKitrick article notes the entire AGW house of cards relies on how one question is answered:
“It would be fine if the climate signal were large and the inhomogeneities were small. But it is the other way around. We are looking for changes measured in tenths or hundredths of a degree per decade, using data from weather stations where the inhomogeneities can easily shift the record by several degrees.”
There is no way a tenth of a degree signal an be detected when the data used has stated error levels of 1-4Â°C! I still don’t understand why this document has not been used to bludgeon the IPCC into admitting it’s conclusions are based on unsupported math.
But there is more damning admissions in the documentÂ (written in 2005)
“We have not yet made final estimates of all the uncertainty components, but we expect the sampling error to dominate the uncertainties of monthly gridded fields. Figure 4 shows the estimated sampling errors for the gridded fields of figure 3″
Figure 4 is the graph above with the huge sampling errors. The bold text says it all. CRU has not proven even to themselves they have the accuracy stated in IPCC literature, based on CRU data. Here are a list of CRU ASSUMPTIONS at the time of writing this document, assumptions which have never been proven as far as I know:
There are three different sorts of uncertainty to be considered:
â€¢ Uncertainties in the station data:
â€“ Measurement error: following  we estimate this as 0.04C on monthly average temperatures.
â€“ Uncertainty in homogenisation corrections: we are estimating this by examining records ofÂ corrections performed at CRU, and by examining differences between corrected and uncorrectedÂ data provided by the Austrian and Canadian national met. services.
â€“ Uncertainty in the climatologies: this is only important where station data is incomplete overÂ the climatology period; we are estimating it by exploring the effect of removing data fromÂ stations with complete coverage.
â€¢ Uncertainties in the gridded fields:
â€“ Effects of the station uncertainties: the combined effect of the uncertainties described above.
â€“ Sampling error: we are adapting the methods of  to work on a flexible grid.
â€¢ Uncertainties in the bias corrections. Following , we are considering two bias corrections: instrumentÂ exposure changes and urbanisation. we are adapting the methods of  to work on a flexibleÂ grid.
Call me crazy, but it looks to me that CRU knew it had not assessed and quantified the error in their data products, and never made that little detail public to all the people who were led to believe the data was accurate enough to detect AGW levels claimed.
Update: Then there are errors created by making up temperatures out of thin air:
The oddity about the picture is that we are given temperature data where none exists. We have very little temperature data for the Arctic Ocean, for example. Yet the GISS map shows radical heating in the Arctic Ocean. How do they do that?
The procedure is one that is laid out in a 1987 paper by Hansen and Lebedeff In that paper, they note that annual temperature changes are well correlated over a large distance, out to 1200 kilometres (~750 miles).
Does anyone really think a temperature reading in Los Angeles, CA can be used to preciselyÂ estimate the temperature in Eugene, OR? I mean anyone old enough to read?