How Bob Carter diddled his data

“We have shown that internal global climate-system variability accounts for at least 80% of the observed global climate variation over the past half-century.”

That is how Bob Carter described the implications of the paper he coauthored last year (McLean et al. 2009). But a new rebuttal to the paper (Foster et al. 2010) describes how McLean et al diddled their data to create a bogus and very strong positive relationship between the ENSO index and global atmospheric temperature. Scientists have long known that ENSO cycles, El Nino and La Nino events, can drive short-term (year to year) fluctuations in global climate and temperature.  But Foster et al McLean et al argued that there was an extremely strong relationship between the two variables, and moreover, that an increase in the ENSO index explained 80% of the observed global warming since, i.e., they argued that ENSO caused global warming.

Foster et al. 2010 is currently in press at the peer-review journal Journal of Geophysical Research but a PDF preprint can be downloaded here. Some aspects of it are technical, but most of the paper is quite readable.

The primary problem Foster et al. 2010 identified in the McLean et al. 2009 is how they transformed or filtered their data before analyzing how related the two variables were.

their [McLean et al] conclusions are seriously in error because their analysis is based on inappropriate application of filters to the data used. It is well established that ENSO accounts for much of the interannual variability in tropospheric temperatures (Trenberth et al. [2002] and references therein). By filtering they have reduced the time series studied to a narrow frequency band, thereby exagerrating what is already well-known. Consequently, their estimates are at marked variance with essentially every other study of the connection between ENSO and large-scale temperature variability, particularly with regard to the role of ENSO in any long-term warming trends, that has been carried out over the past two decades. – Foster et al

It is only because of this faulty analysis that they are able to claim such extremely high correlations. The suggestion in their conclusions that ENSO may be a major contributor to recent trends in global temperature is not supported by their analysis or any physical theory presented in that paper,especially as the analysis method itself eliminates the influence of trends on the purported correlations. – Foster et al

Here Foster et al describe the diddling, in technical terms, that led to the bogus result:

For all monthly time series (the global and tropical MSU temperature estimates from UAH and the SOI from the Australian Government Bureau of Meteorology), the analysis of MFC09 first takes 12-month moving averages of the data, then takes differences between those values which are 12 months apart. The first step filters the high-frequency variation from the time series, while the second step filters low frequency variation. The latter step is perhaps the most problematic aspect of their analysis. It approximates taking the time derivative of the smoothed series, and therefore any linear trend which may be present in the original data will be reduced to an additive constant. Since additive constants have no effect on the correlation between time series, any subsequent correlation-based analysis of the processed time series can tell us absolutely nothing about the presence or causes of trends in the original data. – Foster et al

McLean et al justify the filtering by stating:

“To remove the noise, the absolute values were replaced with derivative values based on variations. Here the derivative is the 12-month running average subtracted from the same average for data 12 months later.”

But as Foster et al point out:

taking the derivative of a time series does not remove, or even reduce, short-term noise. It has the opposite effect, amplifying the noise while attenuating the longerterm changes. Thus, the use of the differencing filter has not been justified, as it has precisely the opposite effect to that invoked by the authors. The noise due to short-term “forces” has already been reduced by the moving-average step. Yet even this noise should not have been removed if the authors truly wish to estimate how much of the total variation in GTTA is due to variations in the SOI.

In spite of the extreme distorting effect of their filter, MFC09 consistently refer to the correlations and fractions of explained variation they derive as between the SOI and tropospheric temperature, both in the abstract and the conclusions. They make no attempt to draw attention to the fact, let alone emphasize, that the reported correlations are between heavily filtered time series, or between estimated derivatives of time series. This failure causes what is essentially a mistaken result to be misinterpreted as a direct relationship between important climate variables. – Foster et al

The second problem with McLean at el is their stitching together of temperature data from two sources in their Figure 7, as a way to suggest their statistical findings are also evident in the raw data trends.  Two aspects of this are fishy.  One, they failed to correct for an offset in one of the datasets, which effectively reduced a recent observed warming trend (see Foster et al’s discussion of this just below).  Two, they effectively hid this stitching wiith vertical lines in their graphic.

In Figure 7 of MFC09, the authors plot actual GTTA (not filtered versions) against the SOI (using different axes) to illustrate the quality of the match between them. However the GTTA signal they plot is a splice of RATPAC-A data through 1979 followed by UAH TLT data since 1980. RATPAC-A data show a pronounced trend over the entire time span, which is visually evident from Figure 4 in MFC09, the temperature line rising away from the SOI line. It is especially misleading simply to append one data set to the other because there is a zero-point difference between the two. The mean values of RATPAC-A and UAH TLT data during their period of overlap differ by nearly 0.2 K, so splicing them together without compensating for this introduces an artificial 0.2-degree temperature drop at the boundary between the two. Unfortunately this is obscured by the fact that the graph is split into different panels precisely at the splicing boundary. – Foster et al

John Cook has a clear explanation of this problem too:

Another interesting feature of McLean et al 2009 is a plot of unfiltered temperature data (GTTA) against the Southern Oscillation Index (SOI) to illustrate the quality of the match between them. However the temperature signal is a splice of weather balloon data (RATPAC-A) to the end of 1979 followed by satellite data (UAH TLT) since 1980. RATPAC-A data show a pronounced warming trend from 1960 to 2008 with the temperature line rising away from the SOI line. This warming trend is obscured by substituting the weather balloon data with satellite data after 1980. It is especially misleading because the mean values of RATPAC-A and UAH TLT data during their period of overlap differ by nearly 0.2 K. Splicing them together introduces an artificial 0.2-degree temperature drop at the boundary between the two. Unfortunately, the splicing is obscured by the fact that the graph is split into different panels precisely at the splicing boundary. This splicing + graph splitting technique is an effective way to “hide the incline” of the warming trend.

McLean et al first author John McLean is an Andrew Bolt palwho often gets Bolt into trouble by sharing misinformation with him. Second author Chris de Freitas has an ethically challenged reputation as well; as an editor at Climate Research he published the notorious (and debunked) Soon and Baliunas paper.  The lead editor and several additional editors at Climate Research resigned over the de Freitas flap (see a round up of this saga here in Scientific American). And third author, Bob Carter is a skeptic media darling, frequently appearing on right wing American talk shows like the Glenn Beck show and speaking at Heartland Institute conferences.

There was already a fair amount of analysis and discussion of the mis-deeds of McLean et al. 2009 even before the Foster et al. 2010 paper was published. Many think some of the anonymous bloggers who first noticed the problems with the study, e.g., Tamino, are indeed authors on the new Foster et al rebuttal paper. Brian Bahnisch has a recent roundup here, including:

Tamino’s explanation of the errors in the analaysis

Greenfyres list of a range of ethical lapses and other problems with the paper

Deep Climates very deep analysis of the problems with the analysis:

Finally, see John Cooks post Foster et al overview of the problems with McLean et al here

References

McLean, J. D., C. R. de Freitas, and R. M. Carter (2009), Influence of the Southern Oscillation on tropospheric temperature, Journal of Geophysical Research, 114, D14104, doi:10.1029/2008JD011637.

Foster, G., J. D. Annan, P. D. Jones, M. E. Mann, J. Renwick, J. Salinger, G. A. Schmidt, K. E. Trenberth. Comment on “Influence of the Southern Oscillation on tropospheric temperature by J. D. McLean, C. R. de Freitas, and R. M. Carter. In Press at Journal of Geophysical Research (download the PDF preprint here)

9 thoughts on “How Bob Carter diddled his data

  1. It’s a well established fact that the anonymous blogger named Tamino is Grant Foster.

    Kind of makes me suspect his motives in pretending to be a third party while boosting his own paper, but nevermind.

    My reason for writing is to ask why do you assume that the a 0.2 rise by the adjusted radiosonde data, which results in a divergence from both the SOI and the UAH, is the natural and correct temperature?

  2. But Foster et al argued that there was an extremely strong relationship between the two variables, and moreover, that an increase in the ENSO index explained 80% of the observed global warming since, i.e., they argued that ENSO caused global warming.

    Suggested correction:

    You probably meant that “McLean et al argued that” etc…

    • Thanks for the link, Marc. Reading now. I’m going to hold off commenting on specifics until AGU give a formal response.

  3. I’m not sure how you can talk about “diddled data” without also talking about Hansen, Hadley/CRU “missing” original temperature data, the Mann “hockey stick” algorithm, the Siberian tree ring fraud, or the admission by NASA that the claimed “independent” NASA GISS temperature data was actually the CRU temperature data since NASA determined their temperature data set was inferior.

  4. Pingback: Climate change and scientific malpractice « Greenfyre’s

  5. Pingback: Lying liar Bob Carter is at it again | SeaMonster

Leave a Reply

Your email address will not be published. Required fields are marked *