Long-time CA visitors will recall the events in mid Sept 2007 when NASA GISS made abrupt changes to US historical temperature data without annotation – a month after the Y2K changes. Some fresh light has been shed on these events by the NASA FOI. At the time, I observed:
no wonder Hansen can’t joust with jesters, when he’s so busy adjusting his adjustments.
To follow the events of the time, look at posts in September 2007. .
I summarized the puzzling changes in a post asking “Should NASA climate accountants adhere to GAAP?”[Generally Accepted Accounting Principles]
On Sep 12, Jerry Brennan wrote in reporting puzzling changes over the previous few days in the station history for Detroit Lakes MN which we’d been using as a test case. I posted up the following illustration of the changes with the commentary shown in the caption:
Figure 1. Original Commentary: Jerry Brennan observed today that Hansen appeared to have already “moved on”, noticing apparent changes in Detroit Lakes and a couple of other sites. Here is a comparison of the Detroit Lakes (combined) as downloaded today, compared to the version downloaded less than 3 weeks ago. As you see, Detroit Lakes became about 0.5 deg C colder in the first part of the 20th century, as compared to the data from a couple of weeks ago.
When I noticed this, I sent an email to NASA notifying them of the effect and asking about it – see here.
A little later, again based on a suggestion from Jerry Brennan, I postulated that the most recent Hansen adjustment to his adjustments came from changing the provenance of his USHCN version once again, with the remarkable corollary that “the temperature increase in Boulder since the 1980s is about 0.5 deg more than they believed only a couple of weeks ago”:
It looks like this is the reason for the conundrum observed in my last post. I never thought of checking to see if Hansen had altered early 20th century values for Detroit Lakes MN between August 25 and Sept 10. It’s hard to keep with NASA adjusters. As noted previously, no wonder Hansen can’t joust with jesters, when he’s so busy adjusting his adjustments.
As a result of revisions made within the last 2 weeks, NASA now believes that the temperature increase in Boulder since the 1980s is about 0.5 deg more than they believed only a couple of weeks ago. Boulder is the home of IPCC Working Group 1, the site of UCAR’s world headquarters, NCAR’s site and home to hundreds, if not thousands of climate scientists. You’d think that they’d have known the temperature in Boulder in the early 1980s to within 0.5 degree. I guess not.
When Hansen capitulated to pressure to release GISS code, I commented here on what I believed to be the relevant interest in temperature records – a comment that seems apt today ( in a CRU context):
Personally, as I’ve said on many occasions, I have little doubt that the late 20th century was warmer than the 19th century. At present, I’m intrigued by the question as to how we know that it’s warmer now than in the 1930s. It seems plausible to me that it is. But how do we know that it is? And why should any scientist think that answering such a question is a “hassle”?
In my first post on the matter, I suggested that Hansen’s most appropriate response was to make his code available promptly and cordially. Since a somewhat embarrassing error had already been identified, I thought that it would be difficult for NASA to completely stonewall the matter regardless of Hansen’s own wishes in the matter. (I hadn’t started an FOI but was going to do so.) Had Hansen done so, if he wished, he could then have included an expression of confidence that the rest of the code did not include material defects. Now he’s had to disclose the code anyway and has done so in a rather graceless way.
I also posed the following small puzzle in respect to the temperature records:
If Hansen says that South America and Africa don’t matter to “global” and thus presumably to Southern Hemisphere temperature change, then it makes one wonder all the more: what does matter?
The FOI records NASA GISS reaction. On Sep 12, 2007, Reto Ruedy wrote to Hansen:
Steve is still having fun with step0, but eventually he’ll find that we did not document how we determined the US brightness numbers we use in our homogeneization.
Checking that step, I noticed that the program that reads Stutzer’s file
is machine dependent, because some stations lie on the edge between 2
cells. If in these cases I use the brighter of the 2 cells the choices
become more robust. 8 of the about 400 rural stations would become
peri-urban, and since we don’t distinguish between urban and peri-urban
stations, these would be the only effective changes.
I looked at the US annual mean series in both cases: they differ by less
than .003 C except between 1880 and 1900; in 1900 the difference new-old is -O.OO5 deg C, in 1880 +.014C .
I would prefer to use add the newer version into the sources. I’d also
like to take the text out of the tar file and make it available
separately – this is probably the only thing that any normal person
might be interested in. The big brightness file should probably also be
stored separately for the convenience of. those who would just want to
look at the programs.
By the way, Steve’s newest “discovery” about the Detroit file is simply
due to our switch from USHCN-1999 to USHCN-2005. He also claims (not in any•email, just in his blogs – so it may not be true), USHCN-2006 is also available; I know that NOAA has it but I can’t find it on the USHCN site.
On 9/12/07, James Hansen «
We should make the explanation about switching to newer USHCN available. But I don’t think that changes should be made that affect the results, even at the O.OOx level, unless/until we discuss it when I am in the office. A machine dependence is not something that we can be assaulted for, but changing the analysis in response to McIntyre adds fuel to the bonfire of his vanities. Jim
Later in the day, Hansen added:
Got Makiko’s phone message. I agree that we need to add a dated statement to a list each time we change the analysis procedure. or input files that are used. This can be a very brief statement. In the present case a simple statement to the effect that we switched to XXX on yyy. If WMO or NCDC changes the data in a file, that is not our change, but when we make a change it should be noted.
The day closed with NASA deciding to post up a backdated change notice, the backdate being prior to my observation of the change, making it look like I’d failed to read what was on their website :
Subject: Re: US brightness index
From: Makiko Sato
Date: Wed, 12 Sep 2007 18:34:21 -0400
To: “James Hansen”
*** What’s New ***
(a) Sept. 10, 2007: The use of the USHCN station records was extended to 2005 from 1999.
(b) Please see “A
Light On Upstairs?” (Aug. 10), “The Real Deal: Usufruct & the Gorilla” (Aug. 16), and
“”Peak Oil” Paper Revised and Temperature Analysis Code” (Sept. 7). for discussions regarding the changes made on August 7, 2007 for 2000-2006 U.S. mean temperatures. ” to my web page called “graphs”, but not on the GISS temperature home page. Are my words OK?
On Sep 17, I observed that their webpage now stated:
September 2007: The year 2000 version of USHCN data was replaced by the current version (with data through 2005). In this newer version, NOAA removed or corrected a number of station records before year 2000. Since these changes included most of the records that failed our quality control checks, we no longer remove any USHCN records. The effect of station removal on analyzed global temperature is very small, as shown by graphs and maps available here.