We have some new Climate e-mails to peruse. This one is from the fourth pdf. I transcribed it by hand so if you find any errors let me know.
Subject: Re: Fwd: US temperature correction graphic and fileIt would be real nice to find out what this is about. At first glance it appears they adjust things in a way that adds almost .4°C to the US trend. Is this a valid adjustment? We would need to go back to the station data and then all the various adjustments and corrections to find out.
Date:Mon, 13 Aug 2007 11:30:24 -0400
, gavin@e-mail, klo@e-mail
I did make those calculations (I assume you mean using only GHCN and hand-adjusting only St. Helens and Lihue, in both cases decreasing the trend, eliminating a 1C and .8C step, resp., as stated in our 1999 paper, also using our urban adjustment.
I only held them back because bringing in a new analysis at this time would confuse the situation beyond hope.
As far as global means are concerned, the effect of our cleaningis slightly negative for the pre-1950 period, slighly positive thereafter, the biggest deviations are -.01C in 1922, +.01C in 2006; the change in 1900-1999 (lin. trend) is .01C/century (i.e. without cleaning it would decrease by .01C).
The US trend however is a different story though not surprising: In addition to the change caused by the UHCN modifications (+.30C for the 1900-1999 change as noted in our 2001 paper: +.14 TOBS, +.16 station hist.adj), the other modifications added .08C/century to the trend. So the trend would decrease by .38C.
The deviations for the individual years caused by the cleaning range from -.13 in 1922 to +.37 in 2006. The optical impression this creates when you look at the table of data is totally misleading: the 1998 anomaly just happens to fall below 1C (.93C) whereas 1921, 1931, 1934 are above 1C (1.27, 1,20, 1.37C) !
Think of it this way though. If the trend has been incorrectly adjusted by nearly .4°C a century then the temperature trend is almost non-existent.
H/T Author and Commenters at Watts Up With That
Cross Posted at Classical Values