Watt about four key charts?

I don’t look at Anthony Watts’s blog, Watts Up With That (WUWT), very often, but I glanced at it today and noticed a guest post called four key charts for a climate change skeptic. Anthony Watts’s pre-amble says

Skeptics often get asked to show why they thinks climate change isn’t a crisis, and why we should not be alarmed about it. These four graphs from Michael David White are handy to use for such a purpose.

However, the rather amazing thing about the post (okay, maybe not that amazing) is that each one of the four key charts is deceptive and misleading.

The first chart purports to show 10000 years of climate change. Not only is it from a single site (an ice core from central Greenland – GISP2) but it is also presented as if it extends all the way up till today. However, the GISP2 ice core data is presented as years before present (BP) and the final data point is for 95 years before present. Also, before present in ice core data is actually relative to 1950, not now. This dataset actually ends in 1855 and so is not only for a single site (which would be expected to show more variability than the whole globe) it doesn’t even show any of our recent warming. If you consider what has happened in central Greenland since the mid-1800s, there appears to have been substantial warming. This Skeptical Science post explains all of this in quite some detail.

The second chart is intended to illustrate that climate models have failed by comparing model projections with observations. However, the chart indicates that we’ve only observed about 0.2oC of warming since 1980, and this is simply not true. Both surface and satellite datasets suggest that we’ve probably had at least 0.6oC of warming since 1980, which actually compares quite well with models. If anyone wants a fair comparison between models and observations there’s Realclimate posts for satellite and surface datasets. The comparison is clearly far better than the chart presented in the WUWT post.

The third chart is rather bizarre as it simply illustrates what happens if you change the y-axis scale. You can make the scale large so as to make the warming appear small, even though nothing has actually changed. The post actually says

To make your point or hide the truth you may change the representation of the data. Both of these charts show the same numbers.

It almost seems as if the author is suggesting playing with the y-axis scale so as to change the appearance of the warming. This seems blatantly dishonest, but maybe one should at least applaud the author for their openness.

The final chart is intended to illustrate the banality of climate change by comparing the last 140 years with the last 10000 years. The implication is that the climate has always changed and that the magnitude of the change we’ve recently experienced is nothing unusual. The comparison, however, is between global temperatures and – again – data from a single ice core in central Greenland. We expect much more variability locally than globally, so the comparison is clearly not fair. Furthermore, both datasets are plotted on the same graph, which makes the modern warming appear very slow in comparison to past changes.

Therefore, these clearly are not key charts for anyone who wants to be genuinely skeptical. If one is genuinely skeptical then one would like to see charts that don’t deceive and misrepresent. It’s one reason why the term skeptical is clearly not appropriate for anyone who promotes such charts. One could put inverted commas around skeptical as that would indicate a form of pseudo-skepticism, but even that seems generous. Other labels are, however, often criticised for being a form of unpleasant namecalling. However, if some are going to knowingly promote misleading charts, maybe they can’t really expect any better. It’s not as if doing so would somehow hamper genuine dialogue since genuine dialogue with those who promote such charts is almost certainly impossible.

Advertisements
This entry was posted in Anthony Watts, Climate change, ClimateBall, Global warming, Research, Science, Watts Up With That and tagged , , , , , , . Bookmark the permalink.

28 Responses to Watt about four key charts?

  1. Nick Stokes says:

    I made similar points to yours in a comment there. On the last point, of comparing global averages to individual sites, I showed a graph of Nuuk, Greenland since 1865 with the GISS land/ocean average (all annual):

  2. Nick,
    Thanks, glad we had a similar response 🙂

  3. I wrote a pseudo paper with lots of charts, which I hope to publish in Scientific American as soon as we force the Baader Meinhof gang to sell it to Trump Climate Publications. A prepublication draft is here

    http://21stcenturysocialcritic.blogspot.com.es/2014/09/a-new-parameter-to-predict-tornado.html

  4. Magma says:

    However, the rather amazing thing about the post (okay, maybe not that amazing) is that each one of the four key charts is deceptive and misleading.

    WUWT deceptive??? Next time, include a warning before dropping a bombshell like that!

    I could have been driving or flying a plane or whatever.

  5. Joshua says:

    ==> …(okay, maybe not that amazing) ==>

    I’ve been struggling as to whether it’s possible for something to be simultaneously fully-expected, commonplace, banal, and amazing.

  6. Magma says:

    A little more seriously, I’m not a regular reader of WUWT but I have the strong impression that over the past ten years the content and participants in both the main posts and comments have shifted noticeably. While older ones often attempted to make technical arguments, however muddled or incorrect, recent years have seen a substantial move towards crankism, blatant data misrepresentation, and right-wing political nuttery.

    An archive of WUWT content over the years would surely provide enough meat for several Ph.D. theses in sociology, psychology or political science.

    And it’s nothing new, but the comments directed at Nick Stokes in this latest WUWT post are fine examples of the Dunning-Kruger effect in action.

  7. Marco says:

    As Tamino and Sou report, in another recent article on WUWT, a professor (yes, he really is one) manages to use a faked Time cover in his story. And of course it is lapped up without many protesting such blatant fakery.

    WUWT was, is and will always be an alternative universe. If anyone wants to understand the brain of Trump, look no further than WUWT. Sometimes it makes sense, most of the times it is batshit crazy.

  8. izen says:

    I think the critique of the graphs is mistaken.
    It treats them as flawed or deceptive descriptions or narratives of the real global climate. But that only makes sense if you start with the a prior assumption that AGW is a real and significant problem.

    These graphs are not intended to convert those that hold that AGW is real. They are in support of the underlying assumption that AGW is at most a minor element of the natural variations. They are effective in re-assuring those who hold that assumption that they are correct.

    I would point out that Nick Stokes graph comparing recent global and Greenland temperatures is actually a good example of a graph that supports that idea. Clearly the local variation and long cyclic patterns completely dominate AGW on a local level. (grin)

    At the risk of getting needlessly semiotic, the argument is not over the accuracy of the suyhet, there is a fundamental difference in the accepted fabula.

  9. A comment on the third chart, both for the WUWT discussion and comments on it here: The vertical scale is not arbitrary, even if it seems it could be. The units ought to be normalized in terms of the long term variability of the data, that is it’s standard deviation, but that standard deviation needs to be calculated in a manner which respects that the series is a dependent data series. So, for example, one way of doing that is a Politis and Romano stationary bootstrap for the data. Simply taking an ordinary standard deviation is incorrect, because it assumes the observations are independent. See the original reference for more.

    When this is done, the vertical axis is presented on a natural scale.

    Using the stationary bootstrap still demands some sensible choices be made, guided by the character of the data, such as the mean block length of the varying block sizes.

    The stationary bootstrap is available with a choice stationary for the value of the parameter b> in the tsbootstrap function of R‘s tseries package.

  10. lerpo says:

    ” If anyone wants a fair comparison between models and observations there’s Realclimate posts”

    Patrick Brown’s page has a comparison of obs vs Hansen’s 1981 projection that’s worth checking out: https://patricktbrown.org/2017/01/18/2016-global-temperature-update-to-hansens-1981-projection/

  11. angech says:

    “each one of the four key charts is deceptive and misleading.”
    The first chart does not show any current date or make any claim about it being current, you may wish to read that into it, but it merely states that 10000 years of climate change produced that graph and clearly states,
    “In all of the 10000 years shown here man played no role in the change of temperature or carbon”

    “the second chart indicates that we’ve only observed about 0.2oC of warming since 1980, and this is simply not true.Both surface and satellite datasets suggest that we’ve probably had at least 0.6oC of warming since 1980,”
    It is labelled as an amalgam of 3 satellite datasets, comes from Roy Spencer and does not show 0.6C. The land sets showed 0.4C over this period
    “All of these yield consistent estimates of the approximate magnitude of global warming, which reached about 0.8°C in 2010, twice the magnitude reported in 1981.”
    and if you took the highest recently 0.6C would be right, for land but that could be seen as a “Clutching at straws GWPF style” argument in this setting.

  12. Marco says:

    “The first chart does not show any current date or make any claim about it being current”

    * x-axis says “thousands of years in the past”, ending on the right with zero. How does angech think people will interpret that zero?

    “It is labelled as an amalgam of 3 satellite datasets, comes from Roy Spencer and does not show 0.6C. The land sets showed 0.4C over this period”

    * [Mod: redacted] Essentially every choice made is clearly made to increase the discrepancy (satellite temps are those of the mid-troposphere known to be ‘polluted’ with stratospheric cooling, satellite temps are offset downward, while they in reality most likely should be moved upward on the y-scale, and no uncertainty is shown). Note also that UAH v5.6 for the lower troposphere shows 0.6 degrees of warming for 1977-2015, and RSS shows 0.5 degrees. HADCRUT4 also shows 0.6 degrees for that period.

  13. Bernard J. says:

    For those who are curious about the WUWT page but are leery of dignifying it with a hit:

    http://archive.is/mnf9o

  14. russellseitz says:

    I tried commending this post to Andrew Montfor’d wights at Bishop Hill, but all I got
    were lean and hungry looks .

    https://vvattsupwiththat.blogspot.com/2017/01/buck-house-balks-at-ukipscience-cut.html

  15. angech says:

    Marco says:
    * [Mod: refers to moderated comment]
    re a graph labelled as an amalgam of 3 satellite datasets,

    John Christy not Roy Spencer, I apologize for that amaurosis but the two are almost interchangeable, Roy gets the most headlines I guess.
    The graph has been shown to be fraudulent?
    With great charges comes great responsibility, Marco.
    Marco would like to move the temperatures up on the y scale.
    You cannot adjust the temperatures just to fit your narrative.
    It would be like me saying the temperatures are fraudulent because they have been contaminated by CO2.
    Showing the actual temperatures is not fraudulent.
    “satellite temps are those of the mid-troposphere” is not fraudulent. He happens to be comparing it with 102 simulations by 32 models of , that’s right, the mid-troposphere.
    Did the models forget to put in stratospheric cooling “pollution”?
    Either they are good models, they put it in as well or the models are not as good as you wish to claim
    [Mod: refers to something moderated]

  16. angech says:

    Marco, The second chart is of the mid-troposphere and this does make a difference to estimating surface warming as you
    “UAH v5.6 for the lower troposphere shows 0.6 degrees of warming for 1977-2015, and RSS shows 0.5 degrees. HADCRUT4 also shows 0.6 degrees for that period.”
    and ATTP
    “Both surface and satellite datasets suggest that we’ve probably had at least 0.6oC of warming since 1980, ”
    are aware when you talk about the warming in different layers or heights
    When ATTP says “the chart indicates that we’ve only observed about 0.2oC of warming since 1980, and this is simply not true.” it is not that the graph is false.
    the reason is possibly that the mid troposphere, being much higher colder and stable does not exhibit as large a range of warming for the same amount of global warming/CO2 rise.
    I felt it showed a 0.28 C increase but either way it is a lot less than 0.6C but right for that height.
    The data set temperature changes are not comparable to each other.
    Aside, if this is right the uncertainty bars would be a lot tighter and smaller than you might wish.

  17. angech,

    the reason is possibly that the mid troposphere, being much higher colder and stable does not exhibit as large a range of warming for the same amount of global warming/CO2 rise.

    Except this is simply not true. Neither RSS nor UAH show as little as 0.2oC of warming between 1980 and now.

  18. BBD says:

    Alternative facts again.

  19. angech says:

    “Except this is simply not true. Neither RSS nor UAH show as little as 0.2oC of warming between 1980 and now.”
    We seem to be on the edge of a controversy over the Christy chart, as I see it features in a Skeptical Science article “How reliable are climate models?”
    where a similar graph shown is labelled global bulk atmospheric temperature and shows a 0.091 C for the 3 satellites and a 0.079 C for balloons per decade. 0.214 for the climate models.
    This would account for a warming detected by the satellites of 0.315 from 1980 approx to 2015 approx for the global bulk atmospheric temperature.
    As I see it the graph you are showing is the TMT (Temperature Middle Troposphere),
    You say 0.02 and I say 0.028 over 35 years and you say this is too low.
    The common surface temperature satellites is the TLT (Temperature Lower Troposphere)
    “UAH v5.6 0.6 C and RSS shows 0.5 C. which is your comment above.
    The one at Skeptical science and Dr Christy’s Senate presentation is the combination global bulk atmospheric temperature 0.315 C from the trend line. suggesting the TMT is indeed lower than the TLT
    I hope my comment on the TMT [shown] being low is correct, a lot of scientists here may confirm this,perhaps Marco may have the figures.
    I am not disagreeing with you. I think we are talking about different graphs and levels. I hope this sorts it out.

  20. angech says:

    BBD says: Alternative facts again.
    No just trying to understand, that is all. Help us if you want.

  21. angech,
    Virtually no temperature datasets today show as little as 0.2 – 0.3oC of warming between 1980 and now.

    I hope my comment on the TMT [shown] being low is correct, a lot of scientists here may confirm this,perhaps Marco may have the figures.

    I don’t think it is correct, at least not in the sense of it being as low as 0.2 – 0.3oC.

  22. Marco says:

    “Marco would like to move the temperatures up on the y scale.
    You cannot adjust the temperatures just to fit your narrative.”

    Congratulations, you actually touched upon one of the problems with Christy’s chart, but you don’t realize it.

    Ask yourself the question why the curves for the satellite averages are BELOW those of the models *for the whole range*. I doubt you have an answer. But Gavin Schmidt has a well-reasoned explanation why they should actually start slightly *higher* than the models:
    http://www.realclimate.org/index.php/archives/2016/05/comparing-models-to-the-satellite-datasets/
    (see also the link to an earlier piece Gavin wrote).

    “Did the models forget to put in stratospheric cooling “pollution”?”

    Yes. They are, by their very nature, a solid representation of the middle troposphere. The satellite temperatures are derived from a complex mathematical model, which attempts to isolate the temperaturse of the different parts of the atmosphere, but likely never fully can do so.

    “perhaps Marco may have the figures”

    See the link to realclimate above. Note in particular how extremely dependent the vaues are on which satellite product you use. That perhaps tells you something. If I recall correctly the discrepancy between the different satellite temperatures is smaller for the TLT product, and the discrepancy with the models is also smaller. In other words, just about every single choice made by Christy manages to inflate the discrepancy with the models. Several of the choices are very difficult to defend (e.g. the baseline issue), others mean little like the smoothing/padding, but again this just happens to artificially inflate the difference a little bit more.

  23. Tony Banton says:

    This tells you all you need to know of the veracity of the tropospheric temperature data, and just where it diverged from reality…..



    RSS acknowledge that either the MSU prior 1997 is in error, or the AMSU after is.
    They take the pragmatic view that they don’t know which and have averaged out the error between them in v4.0.
    UAH meanwhile, have ascribed to the newer AMSU “Cadillac quality”, and maintain that as being correct.

  24. Tony,
    Thanks, I hadn’t seen that before.

  25. BBD says:

    What Tony said.

    Here’s a simple DIY demonstration that UAH v6.x beta is biased cool: compare it directly with RSS TLT 3.3 which has a known cool bias.

    From RSS news release Jan 05 2017:

    RSS TLT version 3.3 contains a known cooling bias. We are working to eliminate the bias in the new version of TLT.

    Reference:
    Mears, C. A. and F. J. Wentz (2016). “Sensitivity of Satellite-Derived Tropospheric Temperature Trends to the Diurnal Cycle Adjustment.” Journal of Climate 29: 3629-3646.
    http://journals.ametsoc.org/doi/10.1175/JCLI-D-15-0744.1

  26. angech says:

    ATTP thank you for that RSS graph TMT supporting your comment.

  27. angech says:

    I notice at RSS Carl Mears states the UAH 6.0 gives a TMT trend of 0.065 C per decade for 1979 – 2015 , this would be 0.23 C rise approx and is one of the 3 satellites used in the graph.
    RSS v 3.3 is 0.08 so 0.28 C.
    If he used version 4.0 the trend of 0.13 would lift the combined trend, his graphs may only be to 2015* hence the older lower version used.
    Virtually no data sets giving as little as 0.30 C from 1980 to now (very true) is quite a different argument to-
    the number of data sets for the TMT from 1980 to 2015.
    Of which there are possibly only 5 complete ones we normally use and 2 of these show less than 0.3 C. That is 40% of the available data.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s