How "Small" an Effect do Adjustments Have?

Hey guys. I've been having trouble finding a topic to write about this week, so I've decided to re-visit an old issue. People familiar with my writing know I have spent a fair amount of time examining what effect human adjustments to temperature data have had on temperature records. This has mostly focused on the work of the (questionably named) BEST group.

I am not going to re-visit the not insignificant history of this topic today. If you want to read a bit about it, this post should give you a bit of an introduction to the matter. There is one historical point I do need to bring up though. A year and a half ago, in April of 2015, the head of the BEST project Richard Muller gave an interview in which he said:

“Furthermore, because of the interest, we re-analyzed all the data with ZERO adjustments, just to see what we would get. These results have been made available online. What we found was that the conclusions we had previously drawn were unchanged. The data are available here

You can read up about the trials and tribulations surrounding that article here if you would like, but the salient point is the data Muller referred to has never been published. It had not been published in April of 2015 when the interview was given, and it has not been published as of today, in August of 2016.

I don't know why Muller claimed data had been published when it had not. I don't know why that data has never been published. That's a matter for another day though. The reason I bring this up today is I want to point out anyone hoping to analyze the effect human adjustments to recorded data have on the BEST temperature results will face the obstacle of the BEST group falsely having claimed to publish the data which would make that possible without completely redoing the BEST analysis.

Fortunately, as you may have noticed while reading the posts I linked to above, the BEST group has shared that data with me. Unfortunately, that only happened after we exchanged a number of e-mails and I publicly criticized them (multiple times) for failing to publish data then turning around and claiming it was published. Still, the result is I have the data. Because of that, I've spent some time comparing the BEST adjusted and unadjusted results. This is what BEST said of them in February of 2015:

The global time series is important, for example, if we want to make estimates of climate sensitivity or if we want to determine how much warmer it is today than in the Little Ice Age or if we want to compare today’s temperature with the temperature in the MWP or Holocene or if we want to make arguments about natural variability or anthropogenic warming.


Figure 1. Unadjusted data results are shown in the blue curve. The green curve shows the results if only metadata breakpoints are considered. The red curve depicts all adjustments.

As Figure 1 illustrates the effect of adjustments on the global time series are tiny in the period after 1900 and small in the period before 1900.

While I was critical of the blog post that was said in at the time, I was unable to do any real examination of what they said due to the fact they had not published the data one would need to check their claims. By the time I obtained any of the data, three months had passed and people had stopped talking about the post. As such, most people were unaware the BEST group's claim "the effect of adjustments on the global time series are... small in the period before 1900" was incredibly misleading.

At the time, I pointed out at the time, the BEST adjustments altered their temperatures in 1850-1900 period by roughly 15-20%. That seemed a bit larger than "small." However, it was only months later I realized the effect was much greater than 15-20% prior to 1850. For some reason, the BEST team chose not to show the 1750-1850 period in their graph. If they had shown it, this is what people would have seen:


As with the BEST figure, the Red line is the final result. The blue line is what would happen if no adjustments were made. The green line is what would happen if only some of the adjustments had been made. To get a better look, let's remove the blue line and smooth the data a bit more:


Here, we see there are differences of over half a degree at some times. That's more than 25% of the change in temperature the graph shows. Would any person reading that blog post have suspected such a large effect when the BEST team said it was "small in the period before 1900"? I don't think so.

And that's just the maximum difference in temperature at any single point. That doesn't account for the fact adjustments sometimes cool the results and other times warm them. Let's get a complete picture of the effect adjustments have:


For a 50 year period, BEST's adjustments warm their results by about half a degree. Then, BEST's adjustments switch to cooling, again by about half a degree. This cooling effect deminishes until about 1940, at which point adjustments have no effect. After 1940, adjustments again warm the results. This final change is smaller than previous ones and tapers off as we reach current times.

BEST's adjustments have a net effect of a degree change in temperatures. That is 50% of the total amount of warming BEST finds in its final results. You wouldn't know this though because BEST chooses not to publish these results anywhere. Even worse, when BEST publishes graphs depicting these results, it cuts off 100 years of its temperature record, hiding much of the effect their adjustments have.

I get the uncertainty in the BEST results prior to 1850 is quite large, but that does not justify going around telling people the effect adjustments have on its results are "small in the period before 1900" while conveniently failing to show 100 of the 150 years worth of results they have before 1900.

Now, the point of discussing this topic is not to criticize BEST. While I think criticism would be appropriate, the topic I wish to discuss is actually a different one. I want to talk about what effect these adjustments have on how people should interpret the global warming issue.

That'll be the subject for my next post though. This post has already run long, and I think it stands on its own well enough. People certainly deserve to know groups like BEST make misleading claims about the effect adjustments have on their results, choose only to display that data which appears to support their misleading claims, choose not to publish the the data which would let people examine the effect of adjustments themselves, then turn around and give interviews in which they falsely claim that data has been published.

As a final note, I should point out the results shown in this post are only for land temperatures. They do not include any data for ocean temperatures. Oceans tend to warm more slowly than land does, so the total amounts of warming will be noticeably different. Other differences may exist as well. I'll discuss that in the next post as well.



  1. Nice piece again Brandon. Just a general question; I understand the need to
    try and 'improve' the signal by re-analysing the raw data. However, I've not
    seen anything convincing on how these reanalyses are validated - have you?

    I generally object to these products being presented as temperature records
    when they are really models of temperature records. None of the graphs I can
    recall seeing (e.g. the GISS graph that Cox recently used) ever have error bars
    or say anything about uncertainty.

    Finally, I don't know if you've ever seen any of Clive Best's work? He did quite
    an interesting piece looking at the effect of temperature records from rapidly developing
    population centres (as a proxy for UHI) on the global trend:

  2. JonA, thanks. Unfortunately, there have been serious issues with pretty much everything I've seen written about temperature records, from both "sides." The thing I find most baffling is how often I've seen results without any published data or code one could use to verify them.

    Even if I accepted the work done to support the validity of temperature reconstructions, I have a number of fundamental issues with the process. The most fundamental one to me is that "global temperature" doesn't have a definition. What it means in any particular case depends on all sorts of choices and details of the methodology being used.

    To see what I mean, consider how land temperatures are usually considered for several feet off the ground, but ocean temperatures are usually considered for underwater measurements. Then, it is often the case parts of the globe are missed or weighted unevenly. There is no coherent definition for what is being "measured" in such a scenario. It gets even worse when you realize a change of temperature in one location doesn't reflect the same thing as a change in temperature in another location. The change in energy for a location on the equator rising by one degree is not the same as the change in energy for a location in the middle of the Rocky Mountains, yet their values are averaged together.

    There are hundreds of details and nuances that go into determining what the "global temperature" will be for a given methodology, and they're never laid out in a clear and precise fashion. Even worse, what gets measured as the "global temperature" changes over time as organizations change their source data and methodologies. That means past results portrayed as the same product are actually measuring different things than current ones. For what's supposed to be a grave threat to humanity, there seems to be little motivation to figure things out precisely.

    A natural result of this is the uncertainty levels given with results never reflect the actual uncertainty. The uncertainty is always understated. The people doing the work usually know this too, but choose not to discuss it in a frank and open manner. I find that shady. I certainly wouldn't tolerate the low standards in behavior and work used by the people making the temperature record if I thought global warming was a serious problem.

    That said, the work done on the other "side" doesn't appear to be any better. I've seen too many people like Clive Best, Ross McKitrick and (shudder) Steven Goddard say things that are simply wrong, if not intentionally misleading. I'm not sure I've seen anyone examine the topic in what I'd consider a trustworthy manner. It seems every discussion I see on the subject comes from somebody trying to prove something they believe, rather than examining things in an objective manner.

  3. Thanks for the response. I agree with most of what you say though I do find Clive Best's work interesting
    as I can more easily follow it (empirical, 'citizen science' for the most part). I've only really become interested
    in this subject since the late last year hence don't have your experience.

    It's probably old hat to you but I'm frankly appalled by the behaviour of certain individuals within the
    professional 'climate sphere' (your friend ATTP being a favourite). Scientific understanding is only every provisional.
    It seems ludicrous to me (having an Oceanography degree and Master's degree in Geophysics) to refuse to debate
    any of the science involved because it's 'settled' and that divergence from the doctrine is 'denial'.

  4. For what it's worth, I haven't read a lot of Clive Best's work. I just happen to remember a couple posts he wrote with significant errors in them, ones that didn't get corrected when they were pointed out (as he disagreed with them with bad arguments). I wouldn't care to try to guess what the overall quality of his work is.

    As for behavior, it's interesting you bring that up. The primary reason I've never found global warming worrying is people's behavior. If I had discovered something I believed to be a serious threat to the future of humanity, I would behave radically different from global warming advocates and climate scientists because I'd want to do everything I could to help humans combat the threat I perceived. That's not what happens. The sheer contempt and rudeness so many people pushing the narrative of dangerous warming is stunning. If you believed global warming were a serious threat, why would you ever behave like that? That sort of behavior could only sabotage your efforts to get meaningful action to happen.

    The people pushing the narrative of dangerous global warming actually prioritize many other things over combating global warming (shocker, people will latch onto causes to pursue their own goals). If they truly believed global warming were a serious threat to humanity, that'd be either irrational or evil. Either way, it doesn't inspire trust.

  5. What trust are you seeking exactly regarding GHG emissions being a long term threat multiplier ?

  6. I'm not sure what you're asking pete best. I don't think I'm seeking anything. I'm just looking at what the effect adjustments on the data have.

  7. for me , as i personally see the notion of the global temperature being known with any degree of accuracy over the timescale mentioned as nothing more than wishful thinking, i will remain of the opinion that every time an animal or human artefact, or tree stump, or plant matter in general , appear from under some previously snow covered area we know at the minimum it was at least as warm as that day in the past in that location. we can start worrying when snow and ice fields disappear without revealing evidence of animals and plants having inhabited the area previously.

  8. The final baseline for the BEST series is something like 1960-1990. I can't remember it offhand, but you can see roughly where it is in the images above because it's the period where the series have the least difference. I'd be able to remember the baseline better except BEST uses multiple baselines for different parts of their process, and they don't lay out choices like that in a clear manner anywhere. You actually have to check their code tofigure out what period they used for one baseline.

    As for the total amount of change in temperature, remember this is a land-only record. It'll show more warming than if ocean data were included. It can make comparing numbers a bit odd. That 1C of warming in the land+ocean record is more like 1.5C in the land record.

    But really, given global warming is said to be a trillion dollar issue, I'd say even a .01C difference should be relevant.

  9. Looks like the total effect is close to 0 with the two series changing position.

    > If they truly believed global warming were a serious threat to humanity, that'd be either irrational or evil. Either way, it doesn't inspire trust.

    Does the 'that' refer to the previous action, or believing global warming is a serious threat?

  10. MikeN:

    Looks like the total effect is close to 0 with the two series changing position.

    The total effect in modern times is certainly close to 0, but that's simply due to the arbitrary choice of baseline. The negative and positive differences prior to ~1900 do largely cancel out, but that still means adjustments to the data have a significant effect for ~150 years of this 250 year record.

    That's all this post is intending to highlight. I intend to discuss whether or not we should care about the differences in that 150 year period in my next post.

    Does the 'that' refer to the previous action, or believing global warming is a serious threat?

    It's referring to the previous action. There's nothing wrong, at least morally, with believing global warming is a threat of whatever magnitude. However, if you believe global warming poses a serious risk to the future of humanity, and you actively insert yourself into the movement which aims at combating it while prioritizing other goals at the cost of combating the risk of global warming, you're either irrational or evil.

    I was just highlighting how horrible it is to believe global warming is a serious risk yet behave in a way which sabotages efforts to combat it so you can push some other cause.

  11. Brandon, I'm surprised that the adjustments are so high in the past compared with more recently. I would have thought the MMTS and TOB adjustments were mainly in the 1960s-1970s, both having warming effect.

    On a separate point, I notice that the 1815 Tambora and 1883 Krakatoa eruptions are clearly visible and almost in proper relative magnitude of deflection to the size of aerosol emission. I am becoming convinced that known volcanic history should be used to validate tree ring and other theoretical temperature proxies since the volcanic record spans a whole millennium.

  12. Ron Graf, the reason adjustments are so great in the early portions of the record is simply that there is little data in the early record. The thing to remember is, despite what the BEST group has claimed, these adjustments aren't designed to correct problems caused by physical phenomena. They're designed to create a homogeneous field. There may be overlap in these two things, but they're not the same thing. You certainly can't interpret specific adjustments as indicating any physical trait.

    On the issue of volcanoes, I think you'll find my next post on this subject interesting.* It discusses that very subject, and it shows an interesting quirk.

    *I've realized I need to stop saying "next post." I have a bad tendency to write posts about things that pop in my head, so promises about the "next post" will often be wrong. In my defense, my original plan was to have the next post on adjustments up on Wednesday, and I think these extra posts in-between are at least a bit interesting.

  13. Also, for an upcoming post, if you have [the time|the inclination] I'd be interested in hearing
    your thoughts on the verification and validation of the these surface temperature realisations.
    Perhaps this has already been covered elsewhere? (it seems like something Steve M or Lucia
    might have talked about) - in fact, I recall reading a piece Zeke Hausfather guest posted on
    Climate Etc. which I found unconvincing.

    As I mentioned in an earlier comment, I understand the motivation for the re-analysis of the
    historic temperature data we have. What I remain unconvinced about is the certainty that what
    we have now is more accurate. The data seem to require regional correlation for the homogenisation
    to be valid but are also treated as i.i.d for the purposes of precision improvements (law of large
    numbers) - all very confusing to a layman such as myself.

Leave a Reply

Your email address will not be published. Required fields are marked *