the Air Vent

Because the world needs another opinion

Santer Update – Another Dead Paper

Posted by Jeff Id on September 28, 2009

Some of you may have noticed that I added Treesfortheforest tothe blogroll this past weekend.  I’m incredibly lazy about these things but I had to add this due to some very impressive and important results by blogger Chad.

There are two blog posts which need to be read on the issue of whether the climate models are accurate, as Santer claims.  There are only four pieces of evidence you need to review in understanding this paper.

First the abstract, note the sentences I’ve bolded.

ABSTRACT: A recent report of the U.S. Climate Change Science Program (CCSP) identified a ‘potentially serious inconsistency’ between modelled and observed trends in tropical lapse rates (Karl et al., 2006). Early versions of satellite and radiosonde datasets suggested that the tropical surface had warmed more than the troposphere, while climate models consistently showed tropospheric amplification of surface warming in response to human-caused increases in well-mixed greenhouse gases (GHGs). We revisit such comparisons here using new observational estimates of surface and tropospheric temperature changes. We find that there is no longer a serious discrepancy between modelled and observed trends in tropicallapse rates.

This emerging reconciliation of models and observations has two primary explanations. First, because of changes in the treatment of buoy and satellite information, new surface temperature datasets yield slightly reduced tropical warming relative to earlier versions. Second, recently developed satellite and radiosonde datasets show larger warming of the tropical lower troposphere. In the case of a new satellite dataset from Remote Sensing Systems (RSS), enhanced warming is due to an improved procedure of adjusting for inter-satellite biases. When the RSS-derived tropospheric temperature trend is compared with four different observed estimates of surface temperature change, the surface warming is invariably amplified in the tropical troposphere, consistent with model results. Even if we use data from a second satellite dataset with smaller tropospheric warming than in RSS, observed tropical lapse rate trends are not significantly different from those in all other model simulations.

Our results contradict a recent claim that all simulated temperature trends in the tropical troposphere and in tropical lapse rates are inconsistent with observations. This claim was based on use of older radiosonde and satellite datasets, and on two methodological errors: the neglect of observational trend uncertainties introduced by interannual climate variability, and application of an inappropriate statistical ‘consistency test’. Copyright  2008 Royal Meteorological Society


You need to realize that Santer achieved his reconciliation of models and temp by eliminating all the temp data and model results since 2000 which is not a minor point as that’s when the well known global cooling trend showed up in even the most heavily corrected datasets.


You need to read Lucia’s recent post which references some amazingly difficult work by Chad at treesfortheforest:


Temperatures of the Tropical Troposphere: Chad brings Santer up to 2008.

Chad’s fantastic work:

AR4 Model Hypothesis Tests

Fourth to get back your sense of humor you need to watch Ben Santer in the Power of POOOOOP!

10 Responses to “Santer Update – Another Dead Paper”

  1. Carl Gullans said

    Wow, both this and the Briffa Relevation, all in one week…

  2. cogito said

    Please, tell me this video with Ben Santer is not real …

  3. Jeff Id said

    It’s real.

    I know…

  4. Kenneth Fritsch said

    I was going to interject that those two Canadians did an analysis similar to that that Chad did and I did one also. Steve M’s and Ross M’s were submitted for publication. I think it is good that all these analyses are performed as they add more information and insight into the differences between instrumental data and model results.

    The Douglass paper had the flaw of ignoring the auto correlation of the residuals and the variation in the measured data (sort of). What was instructive to me was how Gavin Schmidt handled his critic of the Douglass paper. He was attempting to discredit the basic approach of Douglass using an SE in the comparison, while Santer used this approach in his paper. Santer pointed to the flaws of Douglass that I noted above. One could intuitively see from the graphs that Santer presented that differences between instrumental and models abounded. Also Douglass made a break through when Santer acknowledge the use of SE in the the model to instument comparisons – comparisons that Santer, Karl and others had made previously using ranges of model data and then determining whether that wide range encompassed the instrumental data.

    Interesting to see how progress is made in these matters – even when none of the participants will necessarily admit it.

  5. Jeff Id said


    Steve shared with me some of the replies from the review of this submission a long time ago. The team circled the wagons and spit out some fairly um… odd comments. I was rather hoping Steve would choose to share them by now. He may be resubmitting though and not want to piss people off.

  6. timetochooseagain said

    Christy commented on the Santer/Douglass controversy a while back:

    “It didn’t take long for the “consensus” side, which earlier dominated CCSP 1.1 (Karl et al. 2006), to respond. Santer et al. 2008 reconfirmed the numerical results of the question addressed by Douglass et al. 2007. Our question was simply, “When the models and the observations have the same surface temperature trend, do the models and observations agree in the troposphere?” The answer was no. In other words, Santer et al. reproduced the results of Douglass et al. 2007. However, Santer et al. then asked a different question, which might have interest to some, but was not our question as stated above. They asked something like this, “When individual model trends of the surface are allowed to be examined, whether they agree with the observations or not, do upper air trends between models and observations agree?” Not surprisingly, because some individual model trends are quite bizarre, they could answer in the affirmative, but only for models whose surface temperature did not match the observed surface trend. In other words we compared apples to apples and Santer et al. compared apples to oranges. When going back to the fundamental issue of whether models overstate the atmospheric amplification factor, the answer is clearly yes from the observations and models we have. (And in an ironic result, had Santer et al. used UAH satellite data through the most recent year, the models would have failed their test in any case.) In the analysis, Santer et al. used some “old”, “modified” (i.e. SSTs only) and “new” datasets that (a) revealed less surface warming or (b) more upper tropospheric warming. By using these datasets, the apparent discrepancy could be reduced (i.e. cooling the surface or warming up the troposphere in the observations). Then, one unorthodox trick was added – the use of Sea Surface Temperatures (SSTs) only and ignoring the warming of the land temperatures as if they did not matter (which is incredulous since the upper air resides over land too.) Regarding the SST datasets, they used a “new” one –ERSST -which indicated less warming at the surface so when multiplied by the model-calculated factor of 1.3, implies less warming in the upper air – which then was closer to our upper air observations. However, the version of ERSST used in the paper is now obsolete (obsolete trend was +0.076, new trend is now +0.126 °C/decade – 65% warmer!), so the consistency arguments of Santer et al. based on the old ERSST are obsolete as well. The figure below, from Santer et al. 2008 but supplemented with pink comments, is quite complicated, but contains much of the information described herein. This is a diagram of the vertical atmosphere and superimposed are trends for 1979-1999 from various balloon observations and IPCC AR4 model results. The key point here is that the pink cage represents the entire range of model trends under the assumption they produced the observed surface trend (i.e. this gives an apples to apples comparison between models and observations). As can be seen, the observations (brown, red, green, orange lines) lie to the left (cooler) than the coolest of the model trends for the bulk of the lower atmosphere (700 – 400 hPa). Only part of the RICH (red) trends penetrate the cage, though, RICH is influenced by the ERA-40 model forecast scheme which has a clearly demonstrated spurious warming due to improper assimilation of HIRS channel 11 (which renders RAOBCORE v1.2-1.4 obsolete, see below.) The other balloon datasets are not affected by that problem. In another curious avoidance, Santer et al. did not include surface datasets generated by NOAA/NCDC and NASA/GISS to confuse the overall picture again. When these datasets are used (with their higher surface trends pointing to higher upper air trends when multiplied by 1.3), they indeed more closely support the results of Christy et al. 2007 and Douglass et al. 2007 that upper air trends of models and observations are significantly different. Regarding the upper air trend datasets, Santer et al. included RAOBCORE v1.2, v1.3 and v.1.4, which appeared to show a fairly rapidly warming in the upper tropical troposphere (see Fig.) However, the RAOBCORE datasets, which rely on the ERA-40 forecast cycle, have been shown to be spuriously warm in the upper air due to an error in the assimilation of HIRS channel 11 in 1991-2 (noted in earlier papers, but specifically
    identified in Sakamoto and Christy, 2009). Rather, Christy et al. 2007 and Douglass et al.
    2007 used the latest version from the RAOBCORE group – RICH, which was also affected by the spurious warmth in 1991-2 but not as much, and yet found the inconsistency with models was indeed upheld for the layer-average. Again, relying on the various datasets, which have been tested for accuracy, we find no evidence to contradict the results of Christy et al. 2007 and Douglass et al. 2007. (Note the caveat, “which have been tested for accuracy” – papers such as Santer et al. 2008 do no testing, but simply assume that all datasets are equal, such as “new” ERSST or “old” RAOBCORE v1.2, v1.3 and v1.4, and thus ignore the publications which have provided
    the evidence which document significant errors in the ones they prefer.)”

    A key point in this particular issue is the quality of the data sets used, Christy believes that many of the datasets which are “consistent” have been shown to be erroneous for various reasons. The reanalyses in particular suffer from specific errors related to how the data is processed with models.

    A pretty graph:

  7. Kenneth Fritsch said

    Timetochooseagain, I agree with Christy’s comments that you have shown above and in particularly the comment about looking at difference series between the surface and troposphere tropical temperasture trends. That is something that I immediately took into account in my analysis.

    Unfortunately, not completly crossing the t’s and dotting the i’s can get these discussions side tracked.

  8. John M said

    Maybe the pooping video is just part of “talking the talk”.


  9. BillT said

    If it was part of “talking the talk,” you’d think they *wouldn’t* have disabled the comments…

  10. […] I think this makes the same mistake as the much-criticised paper by Douglass et al on model consistency with observations, subsequently excoriated by Santer et al in 2008. Douglass responds to criticism here and a 600 comment discussion follows. Or search on douglass for the saga, see Lucia’s posts and Air Vent. […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: