In the complaint about Swindle by Risk Management Solutions Ltd., endorsed by a gaggle of 37 professors, they allege that the Swindle graphic of 20th century claims in connection with 20th century temperature change contains the following misrepresentations:
Measurements from meteorological stations that have been published by NASA and other agencies show that the there was an overall slight decline in global average temperature between about 1940 and 1976, but much less than that shown on the graph presented in the programme.
this graph does not correspond to any figure for global average temperature that has been published by NASA
(This is one of many claims about misrepresentation in Swindle, but one where I happen to be familiar with the data and in a position to evaluate it. In connection with 20th century temperature change, they also argue that the Swindle failure to discuss aerosols is a “misrepresentation” of a fact or view, an issue which I may return to on another occasion.)
When a corporation, which has contracts with major insurers, makes highly visible public allegations like this, you’d think that they’d go to the trouble to ensure that the claims are correct. Likewise, when 37 climate scientists sign onto the claim, including temperature specialist, Phil Jones, you’d think that they’d also go to the trouble to ensure that the claim is correct. When they go on to argue that communications with the public – into which category their Open Letter to Martin Durkin published on a website surely qualifies – should have proper due diligence, you’d think that they’d double check that a specific claim like this was true.
But hey, this is climate science.
Since the matter is in controversy, I spent some time collecting as many different versions of the Hansen data as I could locate, including a version from Willie Soon used in the Robinson article referred to in the complaint, from which the Swindle graphic appears to have been derived.
First here is the amended graphic referred to in the Complaint, showing a temperature decline from 1940 to the mid-1960s which was discussed in the program by two scientists.
As noted above, Risk Management Solutions and the professors said that no temperature decline of this magnitude occurred in any Hansen global data. For reference, here is Figure 1 from Hansen and Lebedeff 1988, which shows a temperature decline that visually has considerable similarity to the Swindle version. Certainly the similarities are sufficient that you’d think that Risk Management Solutions and the 37 professors would exercise a little caution before making allegations of misrepresentation to a government agency.
In order to check this apparent similarity, I explored a little for versions of the Hansen and Lebedeff data, trying to collect some vintage data in case later versions had “adjusted” the earlier data. Here is a plot of 7 such versions for 1880-1990, one of which (the smoothed version) was the Willie Soon version.
Hansen versions: 1. Manually transcribed version from hard copy listing in Hansen and Lebedeff 1987 ending in 1985; a version ending in 1987 from http://www.stat.unc.edu/faculty/rs/s133/Data/hansen.dat; a version ending in 1994 at http://ingrid.ldeo.columbia.edu/SOURCES/.Indices/.HANSEN/.Global/.Annual/Temperature/T+exch+table-+text+text+skipanyNaN+-table+.html ending in 1994 (with 1951-1980 mean subtracted); a version from ftp://disc1.gsfc.nasa.gov/data/inter_disc/surf_temp_press/old_tmp_dev/giss_glb_ts.txt ending in 1996; a version from
ftp://disc1.gsfc.nasa.gov/data/inter_disc/surf_temp_press/tmp_dev/giss/giss_glb_ts.txt ending in 1998 and the version presently online at http://data.giss.nasa.gov/gistemp/graphs/Fig.A2.txt ending in 2006.
From these versions, I calculated an 11-year running mean and then added 0.4 to match the anomaly-scale of the Swindle graphic (the Soon version is in red.) The Swindle version appears to match the Hansen version archived by Willie Soon up to a change in level (which is irrelevant to the issue of 1940-1965 temperature decline). As someone who’s looked at a lot of data and data versions, my own opinion is that the Soon version is a version of the Hansen data. Thus the claim that the Swindle data is not derived from Hansen data appears false to me. In addition, I would say that they used a “grey” version of the Hansen data, but the use of “grey” versions is endemic in climate science and, if all climate scientists who use grey data were to be disciplined, you’d have the entire Team in the dock for a start.
As to the allegation by Risk Management Solutions and the 37 professors that the decline in the Swindle graphic is “much larger” than in the Hansen “global” data and must be derived from some other series e.g. NH, it is my opinion that the decline in the Swindle graphic is exactly consistent with being obtained from Hansen “global” version and this particular allegation is simply false.
Versions from above figure, after 11-year running mean filter. Note the close match between the Soon version used in Robinson et al and the Hansen and Lebedeff 1987 version (using data transcribed from the hardcopy).
I hope that Risk Management Solutions is more accurate in their day job of consulting for Swiss Re and other major insurers. You’d think that Phil Jones would have been able to recognize a Hansen and Lebedeff graphic. When people endorse due diligence before making public statements, you expect them to have carried out the analysis that I carried out here before making public allegations to an official body. Making false allegations seems more like “risky business” than risk management.
Update: May 7
Some of the RMS comments were about the relationship of the Swindle graph, which we’ve shown to be based on a 1992 version of Hansen and Lebedeff 1988, relative to the 2006 Hansen versions, suggesting that the Hansen and Lebedeff 1988 global version (Swindle) was “more” like the present-day NH version, or even the present-day US version, than it was to the present-day GLB version. In order to evaluate this comment, I’ve re-plotted the data using an 11 year running mean filter, as in Swindle, comparing the HL1988 versions of global and NH to present-day global, NH and US temperatures.
It is readily seen that the mid-century decline in the continental US in current GISS data is much greater than in any of the NH or GLB versions. Thus, the allegation that the HL88 version is “more like” the continental US version is simply a spitball.
If you compare the two GLB versions (red and black), one notices that, during the past 18 years, Hansen has shaved the GLB temperatures in the 1930s and increased them in the mid 1960s, thereby reducing the mid-century decline. While these changes may or may not be meritorious, because Hansen has so many fingers in the pie – adjusting temperature data, making climate models, lobbying for policy changes – the adjustments have not been carried out by an “independent” person. The Hansen adjustments (and similar Jones adjustments) need to be closely examined by someone independent and, until that is done, I think that one should not assign any greater validity to the re-written data than to the original data.
A second interesting point: if you look at the HL88 NH version against the 2006 version (two shades of green), you’ll see much less adjustment in the mid-century period. So the adjustment process has taken place mainly in the Southern Hemisphere and those are the adjustments that an analyst should examine first. While there has been relatively little mid-century adjustment to the NH series, the late 19th century NH values have been adjusted upwards quite considerably. I haven’t studied these adjustments, but these would also need to be looked at in a full analysis.
Back to the RMS letter: is the Hansen version from the early 1990s “more like” the present-day NH version or the present-day GLB version? I’m not sure what the meaning of this sort of beauty contest is, but I’d be hard-pressed to say that Hansen 1988 GLB series is “more like” the Hansen 2006 NH version than the Hansen 2006 GLB version. I think that we can all agree that Hansen has definitely shaved the point spread between the 1930s and the mid-1960s during this period and the validity of this point shaving is the key question here. I’ve also discussed Hansen adjustments to past data here .