Andrew Revkin of the New York Times writes here in a compacted story. Me versus Jor-El. I spent quite a bit of time saying that the errors mattered a lot at the individual station level and were “significant” for U.S. temperature. For example, consider this page at NASA which shows a comparison between temperatures by individual stations for 2000 and after. The majority of values on this graphic are wrong and the entire graphic will have to be replaced.
In an earlier note on this, also cross-posted at Anthony’s blog, I tried to hew my own line between the exaggerated claims in the right-wing blogosphere and the NASA claims of immateriality. Revkin mentioned in passing to me that Gavin Schmidt, while denying the more extreme blogosphere characterizations of the error, had admitted that everything that I had said about the matter had been accurate. Including the diagnosis of the Jor-El Complex, I presume.
Update:As to the claim of Schmidt and Hansen that 0.15 deg C does not “matter”, if they are prepared to stipulate this, then I would submit the graphic below showing U.S. temperature history (new NASA version) since the 1920s showing a trend increase of only 0.21 deg C in the 87 year period.
Source: New NASA Fig D data, August 2007. The t-value for the trend coefficient is only 1.24, which is not 95% significant even for i.i.d. (and there is lots of autocorrelation here.) There is a temperature increase before 1920 which would increase the trend estimate to 0.42 deg C.
Values in the 2000s are elevated and 95% “statistical significance” is not the only relevant metric for data analysis. However, if people are making claims of statistical significance, it’s fair enough to analyze these claims. The issues remaining to the validity of the NOAA and NASA adjustments remain outstanding. What impact do HO-83 hygrothermometers have on this? Has urbanization and microsite effects been properly accounted for – not just in the U.S. record but in Brazil and China and elsewhere? These issues remain outstanding.
Actually, if one plays with this a little, one can even sharpen Gavin’s point a little further. Gavin and Hansen say that a change of 0.15 deg C is not significant. The trend increase in the U.S. from 1930-2006 is 0.13 deg C., which, according to them, is not “significant”. I myself am expressing no views on the matter at this time – I’m simply reporting the implications of their claims.
Revkin on the Hansen Fiasco
Andrew Revkin of the New York Times writes here in a compacted story. Me versus Jor-El. I spent quite a bit of time saying that the errors mattered a lot at the individual station level and were “significant” for U.S. temperature. For example, consider this page at NASA which shows a comparison between temperatures by individual stations for 2000 and after. The majority of values on this graphic are wrong and the entire graphic will have to be replaced.
In an earlier note on this, also cross-posted at Anthony’s blog, I tried to hew my own line between the exaggerated claims in the right-wing blogosphere and the NASA claims of immateriality. Revkin mentioned in passing to me that Gavin Schmidt, while denying the more extreme blogosphere characterizations of the error, had admitted that everything that I had said about the matter had been accurate. Including the diagnosis of the Jor-El Complex, I presume.
Update:As to the claim of Schmidt and Hansen that 0.15 deg C does not “matter”, if they are prepared to stipulate this, then I would submit the graphic below showing U.S. temperature history (new NASA version) since the 1920s showing a trend increase of only 0.21 deg C in the 87 year period.
Source: New NASA Fig D data, August 2007. The t-value for the trend coefficient is only 1.24, which is not 95% significant even for i.i.d. (and there is lots of autocorrelation here.) There is a temperature increase before 1920 which would increase the trend estimate to 0.42 deg C.
Values in the 2000s are elevated and 95% “statistical significance” is not the only relevant metric for data analysis. However, if people are making claims of statistical significance, it’s fair enough to analyze these claims. The issues remaining to the validity of the NOAA and NASA adjustments remain outstanding. What impact do HO-83 hygrothermometers have on this? Has urbanization and microsite effects been properly accounted for – not just in the U.S. record but in Brazil and China and elsewhere? These issues remain outstanding.
Actually, if one plays with this a little, one can even sharpen Gavin’s point a little further. Gavin and Hansen say that a change of 0.15 deg C is not significant. The trend increase in the U.S. from 1930-2006 is 0.13 deg C., which, according to them, is not “significant”. I myself am expressing no views on the matter at this time – I’m simply reporting the implications of their claims.