Eli (RTFR) Rabett, has a new defence of GISS adjustments, arguing that we should simply trust the clergy at GISS. Eli seems to be particularly prickly when it comes to anything that could be construed as criticism of GISS. (BTW Roger Pielke Jr and others have said that Eli is a pseudonym for Josh (RTFR) Halpern, who, among other things, administers summer fellowship applications for Goddard/GISS (http://sffp.gsfc.nasa.gov/faqs.html))
At least, in this case, Eli did not use the perverted persona that he assumed in his criticism of 15-year-old Kristen Byrnes (“Wanna see some pictures lil’ girl?” http://rabett.blogspot.com/2007/07/wanna-see-some-pictures-lil-girl-ethon.html ). Kristen seems to be a modern girl and was rightfully unimpressed by Eli exposing his shortcomings (borrowing David Niven’s memorable phrase), but Eli’s choice of persona was very inappropriate and decidedly unfunny. Eli’s conclusion about GISS was that:
The bottom line is that the ONLY stations which contribute to the overall trend are the RURAL stations
To support this claim, Eli merely quoted several verses from one of Hansen’s epistles, but did not make any independent effort to ascertain how the GISS adjustments actually worked or to replicate GISS/USHCN adjustments or to verify whether it is true that the only rural stations contribute to the trend.
I’m not convinced that this claim is true. Today I’m going to compare adjustments from Tucson U of Arizona and Grand Canyon (an unlit site). It appears to me that the total adjustment process (including USHCN adjustments incorporated by GISS) result in “adjusted” stations becoming a type of blend of urban and rural stations, so that it is not actually correct to say that the overall trend results only from rural stations. At this point, to my knowledge, present methodological descriptions are insufficient to permit an operational replication of either USHCN or GISS adjustments and accordingly I do not have firm conclusions on the matter at the present time. However, as you will see, there is certainly strong evidence that urban trends are affecting USHCN-adjusted rural stations and thus GISS-adjusted values.
First of all, here is a comparison of two different versions of the two series – left, the USHCN TOBS (observation-time adjusted) version for Tucson and Grand Canyon and right, the GISS adjusted version. The USHCN TOBS series for Tucson shows a 4 deg C temperature increase during the 20th century with a pronounced upspike in the late 19th century (which could well be the product of inhomogeneity); on the other hand, the USHCN TOBS Grand Canyon (unlit) series has very elevated values in the 1930s, with recent temperature increases which do not reach levels of the 1930s.
In the GISS adjusted version, the 19th century Tucson upspike has been deleted (not adjusted); this deletion was not done in the USHCN adjustment stage. So even though GISS says that they use USHCN data as is, there is some procedure by which they decided to delete the 19th century upspike. At present, I am unaware of any description of this procedure (but I could have missed it and would welcome any direction to a description). The trend in the Tucson data has been somewhat attenuated, but still remains very strong. The net adjustments in the Grand Canyon series are the more striking: in the USHCN TOBS series, the 1930s are higher than recent values, whereas in the GISS adjusted version, recent values, especially since 2000, are strikingly higher than values in the 1930s. If the trend in GISS data comes from rural stations only and Grand Canyon (and not Tucson) is an unlit station (which I confirmed), how is this possible? I’ll explore the adjustments below.
The two figures below illustrate the adjustment process for Grand Canyon (left) and Tucson parking lot (right). The top panel in each case plots selected versions (USHCN TOBS and adjusted; GISS “raw” and adjusted). GISS “raw” is a misnomer as GISS raw is often similar to USHCN adjusted (but is not necessarily identical.)
The second panel shows the difference between GISS adjusted and GISS “raw” for each series. The GISS adjustment for Tucson is about 1 deg C over the 20th century and is about 0.5 deg C for Grand Canyon – a “rural” site. What is the operational basis for these adjustments – we could, like Eli, mouth the words from the latest reading, but, at this stage, I am unaware of any information which permits an operational replication of these particular values. Also, and this is an important point, this particular adjustment, which does attenuate the Tucson increase, is NOT used in any other gridcell average (to my knowledge). It is my understanding that CRU uses GHCN/USHCN versions – perhaps the adjusted version, perhaps the TOBS version – no one knows. NOAA says that they use the unadjusted version, but I’m not certain that their readme is correct on this point.
The third panel compares the GISS raw to the USHCN adjusted version. As you can see, over most of the history in the two cases, the GISS Raw and USHCN Adjusted versions are virtually identical. This is very important because it means that, although the GISS adjustments do not describe the USHCN adjustments, these adjustments are built into GISS. It looks to me like the USHCN adjustment process feeds back trends from urban sites to adjust rural sites and this contradicts the claim that only rural sites are used in GISS trends. (I’m not 100% sure of this and at this stage am merely presenting graphics.)
There is one really interesting exception to the usual identity of USHCN adjusted and GISS raw series. After 2000, GISS raw takes a 0.5 upward step relative to USHCN adjusted. This is an effect that I’ve seen in other series: after 2000, GISS raw, for some inexplicable reason, appears to pick up USHCN values before TOB adjustment, while after 2000, GISS picks up the USHCN TOBS series. (If the same panel is drawn between GISS raw and USHCN pre-TOBS, there is a virtual identity after 2000.) I can’t think of any valid or even conceivable explanation for this GISS inconsistency – it looks like a programming error: perhaps here at last we have an actual Y2K problem. The impact of this error is not small – it’s a major factor in the jump-up of Grand Canyon values and, in scale, dwarfs any barbecue effect.
The bottom panel compares USHCN adjusted to USHCN TOBS values. Here Tucson has an opposite adjustment to Grand Canyon. These adjustments are supposed to adjust for station moves – the procedure is described in Karl and Williams 1988 [check], but, like so many climate recipes, is a complicated statistical procedure that is not based on statistical procedures known off the island. (That’s not to say that the procedures are necessarily wrong, just that the properties of the procedure are not known to statistical civilization.) When I see this particular outcome of the Karl methodology, my impression is that, net of the pea moving under the thimble, the Grand Canyon values are being blended up and the Tucson values are being blended down. So that while the methodology purports to adjust for station moves, I’m not convinced that the methodology can successfully estimate ex post the impact of numerous station moves and my guess is that it ends up constructing a kind of blended average.
The net impact of this is that, if Grand Canyon neighbors have urban trends, then the USHCN adjusted Grand Canyon version will also incorporate some amount of urban trend.
For reference, here is a table summarizing location changes in the 20th century supposedly embodied in the above USHCN adjustments. with recent recorded station re-locations in 1956, 1967 and 1976. In the 1930s, the station is said to have been at a height of 6890 feet, while the present station (since 1976) is about 115 feet lower at 6785 feet. However, the higher station is adjusted lower by about 1.0 deg C, while the lower station is adjusted up by about 0.4 deg C. Perhaps there is something in the exposure of the stations that justifies this particular adjustment – but this differential USHCN adjustment of about 1.4 deg C to an unlit station is obviously not small.
|Chg||Before (ft)||After (ft)||Before (deg C adjustment)||After (deg C adjustment)|
Here is a similar summary for Tucson (presently in a parking lot at the U of Arizona).
|Chg||Before (ft)||After (ft)||Before (deg C adjustment)||After (deg C adjustment)|