Saturday, February 6, 2010

The Kaufman Correction: Before & After

UPDATE: The "new corrected spreadsheet" from Kaufman 12/09 was just located online and therefore this post revised & updated to reflect the comparison of raw data from the "draft corrected spreadsheet" of 10/09 to the "new corrected spreadsheet" of 12/09 rather than estimating data from the graphs.

Science has published the "Correction & Clarification" to Kaufman et al 2009. Comparing the "draft corrected spreadsheet 10/09"  used to prepare the draft Correction to the "new corrected spreadsheet 12/09" used to prepare the published Correction shows only one difference between draft and published on Record 17:
Update (December 2009)                                                                   
Record 2 was revised using the correct time scale of McKay et al. (S16)                                              
Record 12 was revised to omit the high-pass filter used by Andersen et al. (S25), as was done in subsequent studies (S37)                                    
Record 17 was revised to reflect the interpretation of Geirsdóttir et al. (S29) that temperature is related more strongly to BSi than to the BSi:OM ratio                                         Record 20 was corrected to reflect the interpretation of Tijander et al. (S32) that X-ray density is related inversely to temperature                                                                
Record 21 was corrected to reflect the interpretation of Haltia-Hovi et al. (S33) that varve thickness is related inversely to temperature                                                                   
Records 3 and 10 were revised to correct rounding errors           
Record 19 was truncated at 1799 to exclude the 1805 bin, which was based on only one year (1800)   
A plot of the differences on the sediment proxy temperature anomalies in Record 17 between the old and new spreadsheets shows the old version with significantly higher anomalies especially in the earlier data, and that considerable differences exist between the old and new interpretations of the data:
This change affected the average of the 23 proxies as as follows:  temperature anomaly change from year 0 to year 2000 in the 10/09 spreadsheet=.46 and new spreadsheet=.57, and the corresponding graphs:
Corrected temperature anomaly graph from the draft version:
Corrected temperature anomaly from the published version:
In addition, the highest peak (1.49) was in year 1975 in the draft  data, the highest peak in year 2000 in the published graphs, but 1975=2000 (1.47) in the data contained in the new corrected spreadsheet (12/09).
And here's the dramatic presentation for the public, which improperly tacks on "Mike's Nature Trick" - the instrumental record, shrinks the error bars, and makes the latest "anomaly"~1.2°C for the "trick", sets the 2000 anomaly higher than the 1975 anomaly (even though the 12/09 spreadsheet says they were the same), and adds the backdrop of melting ice for subliminal effect:

1 comment:

  1. Very nice analysis. I've downloaded the Kaufmann excel spread sheet, and you're quite right. The 1975 average is identical to the 2000 average. But the presentation at Arcus shows the year 2000 maximum higher than year 1975. Maybe the need of the narrative trumps the integrity of the result. They presently don't show the melting ice background, though, so could be they thought the better of it after your exposure.