Grilling the Data


9/29/07 UPDATE: We are still waiting on Mr. Steve Bloom to answer this question: “Why is positive bias imparted in USHCN adjustments?”

He incorrectly asserts that he has been “banned” from this blog. Not true. Once he answers this question, that answer along with whatever else he has to say after that will be posted here. Otherwise we’ll continue to wait.

What say you, Mr. Bloom?


Given what NASA GISS has recently done with posting a change to the data methodology on the heels of an error which was embarrasing to them, (see Raising Walhalla) I think this review of a relevant paper might bear some examination:

An Introduced Warming Bias in the USHCN Temperature Database Reference

Balling Jr., R.C. and Idso, C.D. 2002. Analysis of adjustments to the United States Historical Climatology Network (USHCN) temperature database. Geophysical Research Letters 10.1029/2002GL014825.

Abstract and the full paper Download file

What was done:

The authors examined and compared trends among six different temperature databases for the coterminous United States over the period 1930-2000 and/or 1979-2000.

What was learned:

For the period 1930-2000, the RAW or unadjusted USHCN time series revealed a linear cooling of 0.05°C per decade that is statistically significant at the 0.05 level of confidence. The FILNET USHCN time series, on the other hand – which contains adjustments to the RAW dataset designed to deal with biases believed to be introduced by variations in time of observation, the changeover to the new Maximum/Minimum Temperature System (MMTS), station history (including other types of instrument adjustments) and an interpolation scheme for estimating missing data from nearby highly-correlated station records – exhibited an insignificant warming of 0.01°C per decade.

Most interestingly, the difference between the two trends (FILNET-RAW) shows “a nearly monotonic, and highly statistically significant, increase of over 0.05°C per decade.” With respect to the 1979-2000 period, the authors say that “even at this relatively short time scale, the difference between the RAW and FILNET trends is highly significant (0.0001 level of confidence).” Over both time periods, they also find that “the trends in the unadjusted temperature records [RAW] are not different from the trends of the independent satellite-based lower-tropospheric temperature record or from the trend of the balloon-based near-surface measurements.”

What it means:

In the words of the authors, the adjustments that are being made to the raw USHCN temperature data “are producing a statistically significant, but spurious, warming trend in the USHCN temperature database.” In fact, they note that “the adjustments to the RAW record result in a significant warming signal in the record that approximates the widely-publicized 0.50°C increase in global temperatures over the past century.” It would thus appear that in this particular case of “data-doctoring,” the cure is worse than the disease. In fact, it would appear that the cure IS the disease.

From the paper: Our analyses of this difference are in complete agreement with Hansen et al. [2001]

and reveal that virtually all of this difference can be traced to the adjustment for the time of observation bias. Hansen et al. [2001] and Karl et al. [1986]

The reviewer notes: “Our prescription for wellness? Withhold the host of medications being given and the patient’s fever will subside.”

Originally from CO2Science


23 thoughts on “Grilling the Data

  1. Interesting post. I notice the paper was published in 2002, why hasn’t it received more attention? I can’t find any rebuttals/critiques of it on RC, Tamino or Rabett’s sites.

  2. Chris, Google Scholar is a free resource for checking papers for future citations. It would have been helpful for Anthony to have done that first himself, in which case he would have found out that this paper has been largely ignored in the literature. Keim et al seem to have cited it only in order to dismiss it: “One could argue that these differences may be spuriously induced by the adjustments made to the USHCN data as proposed by Balling and Idso (2002). However, these adjustments all have a sound empirical foundation based on the literature cited above.” (p. 4 ). The CO2Science website is notorious for not updating its discussions of papers like this one.
    The general issue of USHCN trends and adjustments has gotten lots of scientifc attention recently, so it’s not as if the general issue is being ignored.

  3. The conclusion of Balling and Idso is juxtaposed nicely with your September 10th posting, specifically the plot of GHCN stations. I believe it lends an important perspective to the United States 1.54 percent of Earth’s surface area.

  4. Steve Bloom –
    The question, as usual, is not whether other people have mentioned it to dismiss it, the question is, whether it is right.
    Now, what do you think? Are their statistical claims about the series right or wrong?

  5. Idso is co-author. Anything associated with Idso is automatically flagged as oil-supported/denialist/etc, by the RC folks, Tamino, and Rabett.
    It boils down to whether or not the TOBS adjustments are valid and accurate.

  6. (My old Lefty self) Why, Michael J, you must know that the Idsos are wild desert crazies who live in that horrid Red State, Arizona. Why, that’s where Goldwater came from! I don’t trust anything from Arizona. Also, Idsos DENY that CO2 is a toxic, deadly poison, worse than Plutonium! They actually …. hahah … refer to it as …. FERTILIZER! But even that is no good, unless it’s ORGANNNNNIC fertilizer, like what I saw living in Bali (/My old Lefty self)
    There is a very memorable passage in the book “Ecotopia Emerging” where the father of the future president of Ecotopia, with her in tow on a backpacking trip, stops at the Sierra Crest, looks out to the East over the Basin and Range section of Eastern California and Nevada, and remarks, to the effect – “Desert crazies…. they’re desert crazies out there. Tearing up the land….” blah, blah, blah. This diatribe has a measurable effect on the future pres. She later goes on to lead the Ecotopian insurgency against the Federal Government and becomes Ecotopia’s first, doctrinaire, highly charismatic leader, in the tradition of Stalin, Hitler, Mao, Castro, Guevara, Chavez, Ortega, etc.

  7. What I really want is the basic relevant facts.
    From 1930-2000
    –GISS raw data shows a mean -.04 cooling per decade .
    –GISS adjustments are +.05C per decade. This is a spurious adjustment. Overall adjusted result: +0.6C per decade.
    Question: Is this all the raw data or just the data of the type-1 and 2 stations?
    –Adjustments are said to have been made for (some) UHI, using “night-light” methodology.
    But no adjustment has been made for the glaringly obvious microsite violations other than the above.
    Do I have this straight?
    I have read that the adjustment made to satellite data is quite correct and compensates for orbital shift. But I think that “raw data” could use a look-see, too, as to how NASA adjusts it.
    Quien Sabe? There may be an Atmospheric Walhalla out there just waiting to happen.

  8. Karl’s TOBs adjustment–adjustment for changes in Time of Observation– is based on a Model..
    Here is a link to Karl’s paper:
    Some facts.
    1 Karl Studied 79 stations across the US.
    2. Karl validated his model by comparing it to 28 stations.
    3. The model is an ESTIMATE with a substantial error.
    That error is multidimensional based on
    1. Latitude
    2. Longitude.
    3. Season.
    4. Cloudiness
    5. Day of the month
    There is a simple question here. In Representing the temperature of the US NOAA and NASA use 1221 stations.
    Those stations have had their data “corrected”
    The correction models use 28 stations to validate themselves.
    So, 28 stations is enough to validate a “model”, the TOBS model, which makes adjustments up to 2C.

  9. Hmm, my previous comment disappeared. Trying again:
    The paper’s conclusions are based on its findings that the UAH MSU and SONDE datasets show more cooling than the three that are actually used:
    “We certainly realize that the conterminous United States represents only 1.54 percent of the Earth’s surface area, and analyses of that areal unit may have limited interpretations for any global temperature record. Nonetheless, we show clearly that adjustments made to the USHCN produce highly significant warming trends at various temporal scales. We find that the trends in the unadjusted temperature records are not different from the trends of the independent satellite-based lower-tropospheric temperature record or from the trend of the balloon-based near-surface measurements. Given that no substantial time of observation bias would be contained in either the satellite-based or balloon-based measurements, and given that the time of observation bias is the dominant adjustment in the USHCN database, our results strongly suggest that the present set of adjustments spuriously increase the long-term trend.”
    The problem is that problems were suspected with both of those datasets in 2002, and subsequently those suspicions were confirmed (follow the links to the prior extended discussions; see also here). So in this case the lack of citations was a result of the paper being deemed of little value as soon as it saw print.

  10. Take Steve Bloom’s post with a grain of salt where he says “So in this case the lack of citations was a result of the paper being deemed of little value as soon as it saw print.”
    Minimizing opposing views and papers to AGW is a common theme for Bloom and many other like him. For example, many opposing views are never posted on the blog
    This paper was using data believed valid at the time, the data was modified through discovery later, much like Hansens recent revise to fix his GISS error, which lessened the impact of this paper, but did not negate the value of the analysis.
    Steve McIntyre is demonstrating on that adjustments may add bias to the temperature record.

  11. I think you missed an important point about this particular paper, Anthony: It didn’t say anything very interesting. That there was a conflict between the various datasets was widely known at the time. The only novelty of the Balling and Idso paper was to calculate that *if* it was the case that the UAH MSU and SONDE datasets were valid, the only thing that could explain the discrepancy with the other datasets was the TOBS adjustment. In effect, they quantified something that everyone already knew about. So even if it had turned out later on the the UAH MSU and SONDE datasets had no problems, this paper would still not have gotten much attention. That most of the people in the field already suspected there was something wrong with those two datasets obviously didn’t help.
    But let’s also not forget the other lessons here: CO2Science isn’t very good about keeping these references up to date, and Google Scholar is an excellent tool for checking up on them.

  12. Bloom I think you are the one missing the point.
    Adjustments are the issue, and it is becoming clear that TOB introduces a bias of its own, so do others.
    From John Langs post on CA:
    You can see how the different versions relate to one another in this chart.
    In terms of how each version’s adjustments add to the overall trend since 1900 in temperature:
    – TOBS adds 0.35F to the RAW data
    – MMTS then adds 0.05F to TOBS
    – SHAP then adds 0.25F to MMTS
    – FILNET then adds 0.1F to SHAP
    – FINAL reduces 0.1F from FILNET

    That more positive bias than negative, and the chart is from NCDC, so you can’t label it anything but what it is.
    So let’s hear your explanation as to why so much positive bias has to be added to the dataset. I’ll give you a freebie, we know why MMTS bias exists, but given the results of my study, the number looks low.
    But I’m not going to fall into the endless argument trap like you loved to set on Pielke’s blog, which is part of why the poor guy wore out and decided to give up.

  13. All of which has absolutely nothing to do with the Balling and Idso paper since they didn’t engage in any such analysis. You should make a correction to the post, but I’m sure you won’t.

  14. Is this 0.65C warm bias introduced throughout, or isit “start on one end and end on the other” deals?
    (And in any case doesn’t the MoE kind of stack up after so many adjustments?)

  15. The paper was about adjustments. The point is they looked at TOBS adjustment, comparing it to an available dataset at the time. The dataset changed, much like the flavor of the week at GISS. But by NOAA’s own analysis, TOBS adds 0.35F to the RAW data, irregardless of the existence of MSU and Sonde or the deltas between them and USHCN. The point is that adjustments have added a positive bias, the question is why.
    Steve Bloom still hasn’t answered that most basic question, and I suspect he’ll sidestep it just as he did in the post above.
    Let’s here it Bloom, why must adjustments add a positive bias to the surface record?

  16. Now we’re onto a different subject. I officially give up on your making a correction to this post, although I would point out that whatever moral authority you might have claimed as an “auditor” has now gone up in smoke. But enough of beating that dead horse.
    Tell you what, though: You do a post explaining wht you think the TOBS adjustment is problematic and I’ll promise to do a serious response.

  17. Ah, Mr. Bloom.
    Your posted comments have shown whatever corrections you envision. You’ve been given ample opportunity to show your concerns.
    The majority of the post is from CO2Science, and attributed. The others are from AGU. The papers themselves I cannot change. Are suggesting that I rewrite their articles to suit your views?
    And I’ll point out that you didn’t mentioned “correcting the post” until the 2nd to last comment you posted. All your previous ones were of the “the paper didn’t get much attention” flavor. You switched from “paper got little attention” to “correction” late in the game, yet you say I’m the one with the credibility problem.
    See here is the thing, clearly from your posts here and elsewhere, you’ve always held the work that I and others have done in this area in contempt. So I’m not in the least concerned that your opinion of me has “changed” when its simply more business as usual.
    Yes the MSU/SONDE data changed, but the paper and article’s take on USHCN adjustments is still valid becuase it is being seen in other investigations today.
    I have two sources of info that say there’s an issue with USHCN adjustments. The Balling paper and NOAA’s chart. We’ve also seen recent evidence that GISS adjustments caused errors, and now subsequent unannounced adjustments have changed the data again after its last “correction”. So that makes three. Then we have one thing you note saying because the MSU and Sonde data was corrected, that invalidates the Balling paper.
    Ok to humor you, lets just say that’s true. Pfft! Balling/Idso paper is gone. Now we have the NOAA chart and the recent GISS discoveries, plus some of the points made by other commenters.
    Ok, 2 out of 3 remain, same question: Why is positive bias imparted in USHCN adjustments?
    Your answer to my question of “why” was “quid pro quo” rather than address the issue head on. You’ve used the old two step shuffle for the disengenuous, i.e. answer a challenge by putting up another.
    I had already planned to do a look at TOBS and other adjustments in the future. The only difference is that you won’t be commenting on them because until you honestly answer my question on USHCN adjustments and bias posited prior to your diversion, you won’t have posting priviledges.
    Ever since Pielke closed shop you’ve been camped out here, squatters rights are not a guarantee of permanent residence. You’ll have to do better than quid pro quo.

  18. Ah. The graph the Rev linked to is in F, not C. So there’s 0.65F upward adjustment.
    And it DOES start at or near zero in 1900 and rise to 0.65 at the present.
    As to that Heraclanian leap in TOBS adjustment since 1980 . . .
    Well, that accounts for half the difference between NOAA adjusted (adjusted up .65F, shows increace of .7C) and NASA raw data (shows insignificant decrease of Ah. The graph the Rev linked to is in F, not C. So there’s 0.65F upward adjustment.
    And it DOES start at or near zero in 1900 and rise to 0.65 at the present.
    As to that Heraclanian leap in TOBS adjustment since 1980 . . .
    Well, that accounts for half the difference between NOAA adjusted (adjusted up .65F, shows increace of .7C) and NASA raw data (shows insignificant decrease of < 0.1C).
    One wonders how to account for the other half.

  19. BTW, Rev, is the “TOBS 3.5” thing a conscious adjustment being made with at least some logic behind it?
    Or is an unjustified, unaccounted for bias that turned up because NOAA was too busy mixing apples and oranges?

  20. Evan, since Mr. Bloom still hasn’t answered the question posed to him (place your bets) I’ll address the TOB adjustment in detail in a future post.
    In the meantime, a thread on the TOB adjustment has been started over at CA:
    BTW, Bloom’s complaining on Rebetts Blog that he is “banned” from this blog, which isn’t the case at all. Banned implies permanence with no options for redemption. Bloom is simply on hold.
    In Bloom’s case he has an option. All he has to do is answer the question posed to him (before his quid pro quo diversion tactic)about why there is predominantly positive bias in post measurement adjustments to the USHCN data. Onc ehe does that he’ll go off “hold”.
    Mr. Bloom, what say you?

  21. Pingback: Litigious Lunacy « Watts Up With That?

  22. Pingback: Barbecues Climate Stations: Reality Catches Up With Fiction « The Unbearable Nakedness of CLIMATE CHANGE

Comments are closed.