Skip to content

Failing Computer Models

January 21, 2021

By Paul Homewood

  

If anybody tries to tell you that the computer models are accurately predicting global warming, show them this:

image

http://www.remss.com/research/climate/#:~:text=The%20RSS%20merged%20lower%20stratospheric%20temperature%20data%20product,in%20well-mixed%20greenhouse%20gases%20causes%20by%20human%20activity. 

 

 It comes from RSS, who monitor atmospheric temperatures via satellite observation. They are ardent warmists, and here us what they have to say:

Over the past decade, we have been collaborating with Ben Santer at LLNL (along with numerous other investigators) to compare our tropospheric results with the predictions of climate models.  Our results can be summarized as follows:

  • Over the past 35 years, the troposphere has warmed significantly.  The global average temperature has risen at an average rate of about 0.18 degrees Kelvin per decade (0.32 degrees F per decade).
  • Climate models cannot explain this warming if human-caused increases in greenhouse gases are not included as input to the model simulation.
  • The spatial pattern of warming is consistent with human-induced warming.  See Santer et al 2008, 2009, 2011, and 2012 for more about the detection and attribution of human induced changes in atmospheric temperature using MSU/AMSU data.

But….

  • The troposphere has not warmed quite as fast as most climate models predict.  Note that this problem has been reduced by the large 2015-2106 El Nino Event, and the updated version of the RSS tropospheric datasets.

It is of course nonsensical to argue that a record El Nino, which raised global temperatures by 0.7C, should in any way be factored into the appraisal of model accuracy.

Take that out, and the models are clearly failing. Actual temperatures have consistently been trundling along the bottom of that yellow band, and more often than not below it, for the last decade or more.

What makes this failure even more damning is that there was a reasonable fit up to 2005, when the models were forced with historical values of greenhouse gases, volcanic aerosols, and solar output ( a process known as backcasting). There is of course nothing clever about tweaking your models until they give you the result you are looking for. But that does not mean the models have any value in forecasting, as subsequent events have shown.

For the actual data to have diverged so much in the space of just a decade suggests the models are worthless.

And we must not forget this comment by RSS:

Note that this problem has been reduced by the large 2015-2106 El Nino Event, and the updated version of the RSS tropospheric datasets”

 

You may recall that in 2016 RSS suddenly dropped their highly inconvenient dataset, below, which inconveniently showed that global temperatures had fallen since 1998:

trend

https://notalotofpeopleknowthat.wordpress.com/2016/03/04/rss-nobbled/

 

In favour of the new version, which showed the opposite:

 

https://woodfortrees.org/plot/rss

.

When the models don’t even agree with tampered data, you have big problems!!

34 Comments
  1. Broadlands permalink
    January 21, 2021 2:24 pm

    And we must not forget this comment by RSS: “Note that this problem has been reduced by the large 2015-2016 El Nino Event.”

    And don’t forget that all El Nino events are natural events that have not been affected by human additions of CO2. There is no correlation between the three-part ENSO and Mauna Loa CO2. A model should not use El-Nino 3.4 indices without the other two…”El-Nada” and La-Nina.

  2. terryfwall permalink
    January 21, 2021 2:25 pm

    Is anyone creating computer models of climate trends that do NOT deliberately set out to prove that human emissions must be the cause of any warming that occurs? If those gave an explanation for temperature change excluding man-made causes it would provide a powerful counter-argument. It would make a change for the climate establishment to have to comment on models that demonstrated a different set of results.

    If this showed that there was only a minor effect, then all the efforts currently being spent on this small part of just one topic could be re-directed to address the many areas in which humans are genuinely despoiling the planet and get something done about those.

  3. It doesn't add up... permalink
    January 21, 2021 2:35 pm

    Surely a good model would predict El Niño events?

    • dearieme permalink
      January 21, 2021 5:10 pm

      Spot on, my son.

      As a former mathematical modeller may I just say that the absurd claims made by these climateers, and the contemptible efforts of the Astrologer-Royal to predict deaths by Covid, leave me feeling a warm glow of satisfaction at my own work.

      That’s probably not good for my character but I’m too old to care.

    • Broadlands permalink
      January 21, 2021 5:30 pm

      That’s precisely why there are no good long-term models. Natural variability is always unpredictable, volcanoes, earthquakes, jet streams. El-Nino events are even difficult to predict, short-term…

      “Shades of Chaos: Lessons Learned About Lessons Learned About Forecasting El Nino and Its Impacts”. Michael H. Glantz 2015

      • Phoenix44 permalink
        January 22, 2021 9:56 am

        Natural variability is wholly predictable – if you understand the mechanisms! But we don’t and as the variability is much greater than any supposed warming signal it’s impossible to determine signal from variability. That us why the claims are false – you cannot claim to know the CO2 warming if you don’t know the natural warming.

  4. MrGrimNasty permalink
    January 21, 2021 3:24 pm

    Interesting that they describe it as a ‘problem’ [finding enough warming via data-tampering/confirmation bias adjustments] and not an error with the theory/models!

    • Harry Passfield permalink
      January 21, 2021 6:37 pm

      Very good point.

  5. Curious George permalink
    January 21, 2021 3:56 pm

    The graph looks funny: Why “global (70S to 80N)”? And it shows that the modelled TLT Anomaly (whatever it is) grows faster than the RSS dataset. So what?

    • January 21, 2021 5:22 pm

      Satellites cannot get readings over the poles – UAH have the same problem

      Mind you, there are virtually no surface readings up there either!!

      The yellow band is what the models are predicting, which is way above actual

      • bobn permalink
        January 21, 2021 6:22 pm

        The RSS temp is computed and not ‘actual measurements’. As a result it has probability error bars. Previously RSS showed this temp range with its high and low 3std deviation error parameters. Now RSS is shown as the single line on the HIGH std deviation line. RSS is no-longer a good measure, however i believe UAH is still trying to stay honest, though under extreme pressure to overheat its results.

      • January 21, 2021 9:39 pm

        Quite right Bob!

      • Peter permalink
        January 22, 2021 2:54 am

        “Mind you, there are virtually no surface readings up there either!!”

        Are there any reliable readings in Africa and South America? The most dense networks are in the USA and Europe. These continents cover less than 10% of the earth’s surface.

      • January 22, 2021 9:59 am

        Very few

      • Curious George permalink
        January 22, 2021 3:51 pm

        Paul, my problem is with 70S-80N. Why not 70S-70N? Or 80S-80N? Would it show something undesirable?

      • January 22, 2021 6:30 pm

        I understand that the high altitude of Antarctica alsois problematic

  6. January 21, 2021 4:54 pm

    The models are full of false assumptions (such as clouds) which guarantee warming as CO2 increases. We all remember what the IPCC said about climate models: “In climate research and modeling, we should realize that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”

  7. ThinkingScientist permalink
    January 21, 2021 7:38 pm

    GCM’s cannot predict El Nino’s therefore, as PH points out, any model fit relying on El Nino’s to get them inside the error bars is a failed model. Because of this issue, I have been working on a method to reliably estimate and subtract the signature of El Nino’s from the global temperature series, so underlying trends can be examined more closely. The method works and also gives an estimate of the El Nino signature. I am currently planning to work this up into a paper.

  8. tom0mason permalink
    January 21, 2021 8:39 pm

    And another point to note —
    As one of the top climate scientists in the world, Kevin Trenberth said in journal Nature (“Predictions of Climate”) about climate models in 2007:

    None of the models used by the IPCC are initialized to the observed state and none of the climate states in the models correspond even remotely to the current observed climate. In particular, the state of the oceans, sea ice and soil moisture has no relationship to the observed state at any recent time in any of the IPCC models. There is neither an El Nino sequence nor any Pacific Decadal Oscillation that replicates the recent past; yet these are critical modes of variability that affect Pacific rim countries and beyond. The Atlantic Multidecadal Oscillation, that may depend on the thermohaline circulation and thus oceanic currents in the Atlantic, is not set up to match today’s state, but it is a critical component of the Atlantic hurricanes and it undoubtedly affects forecasts for the next decade from Brazil to Europe. Moreover, the starting climate state in serveral of the models may depart significantly from the real climate owing to model errors. I postulate that regional climate change is impossible to deal with properly unless the models are initialized.
    ¯

    ¯
    Therefore the problem of overcoming this shortcoming, and facing up to initializing climate models means not only obtaining sufficiently reliable observations of all aspects of the climate system, but also overcoming model biases. So this is a major challenge.

    And as far as I’m aware this is the case today.

  9. John189 permalink
    January 21, 2021 9:16 pm

    I think that many modellers fall into a philosphical, or perhaps better, a linguistic trap. Models are projections developed from input. Beware GIGO. But more importantly, models are not fact. They demonstrate what might be, not what will be. They do not “show us” anything; they merely flag up possibilities. As such they can have their uses, but they will always fail if they are skewed to prove what they set out to discover. Which sadly is often the case.

    • Phoenix44 permalink
      January 22, 2021 9:53 am

      The basic issue is that models model your assumptions. That’s fine, that’s useful. But climate modellers – and Covid modellers – believe they are modelling reality and that their model has “emergent properties” that can e.g. tell them what ECS is.

      This is simply nonsense. The model are just not that good, partly because our understanding of the basic principles is lacking, partly because models of complex, non-linear systems can never be particularly accurate and partly because such systems are extremely sensitive to starting conditions which we never know well enough.

  10. Bruce of Newcastle permalink
    January 21, 2021 9:33 pm

    The Russian model fits UAH pretty well.

    Of course it just happens to have the lowest value for climate sensitivity of all the models. And a much bigger inertia for the oceans, which I think may allow it to emulate the thermohaline cycle – which accounted for a third of warming last century due to the cycle being at bottom at the start of the century and at peak in 2000.

    No wonder the Russians don’t seem to be getting excited about climate change. They appear to have actual climate scientists not fake ones.

    • MrGrimNasty permalink
      January 22, 2021 9:56 am

      Putin has made contradictory statements over the years on climate change but mostly disparaging, he has said it was natural, cyclic, started before emissions could have caused it, his best scientists had decided it isn’t an issue, etc.

      But he seems to have made a political decision to ‘go along’ with it. As evidenced by his funding/propaganda interference in the UK with anti-fracking groups, it is in his interest to watch the West (enter Biden!) destroy its fossil fuel industry and become dependent on his gas at his price – they’ve recently announced more big gas/hydrate finds in Siberia.

    • dave permalink
      January 22, 2021 12:41 pm

      Note that the Chart from RSS is way out of date. It seems (badly drawn axes) to end in 2017. The three or four years since have been flat-to-down for both RSS and UAH.

      Meanwhile the Sun is quiet.

      A few years on from two dubious ‘resets,’ – by RSS for temperature data, and by SILSO for sunspot data – and the cracks are showing through the cowboy paint jobs.

  11. January 22, 2021 12:27 pm

    Computer models are only as “good” as what is put into them. Sometimes we don’t know what we don’t know and therefore don’t know everything to put into them. As we have seen in recent times, they are also subject to and used for an agenda. We have seen computer models used as “truth” over the actual available data to the contrary.

    • dave permalink
      January 22, 2021 12:51 pm

      ‘Computer models’ is misleading talk. They are simulations in computers of the possible evolution of short-term causal models into a comparatively distant future. Extrapolation to the nth degree.
      ,
      And that is quite different from having a robust model, of any kind, of (as opposed to for) the long-term

      • Gamecock permalink
        January 24, 2021 6:57 pm

        Yes, dave. Their use of the word ‘model’ is bogus and should be stopped.

  12. ThinkingScientist permalink
    January 22, 2021 5:12 pm

    For those who don’t know, irrespective of the physics in climate models they only appear to work because they are driven by the input forcings. Without the input forcing curves the models are just noise generators.

    Irrespective of whether the physics is correct in climate models, the trends in temperature that they generate are only there because they are in the input forcings. This means that the climate response output is purely due to the input from the forcings. Any other variation in output is instability or noise.

    So when the modellers claim that if they don’t put human forcing in, the models don’t reproduce historical temp series, they are not proving anything other than the input forcings are already matching the temp series by simple linear transform a priori. Ie the match of the output is obtained purely because the assumed input forcings already give the answer required.

    The climate models themselves may add detail eg latitude variation, land/ocean etc but that can be done easily in a spreadsheet. The climate models are basically lipstick on the pig – the pig being the self-fulfilling input forcings. Without them, climate models don’t do anything. Give the models different forcing’s and they will give different outputs.

  13. M E Emberson permalink
    January 23, 2021 5:11 am

    https://psychology.wikia.org/wiki/Extrapolation

    I thought I would look up extrapolation . It is often used in “prehistory” of a sort. That is, “this culture has crouched burials with sea shells added .. therefore similar cultures with burials are connected “. I don’t need to point out extrapolations concerning pyramidal structures. the newspapers are full of them.
    This psychology study maybe of use to some of us. But don’t let me extrapolate from my own interests to the interests of the real scientists among you. !

    • dave permalink
      January 24, 2021 11:06 am

      Another analogy I like is that tweaking models when they disappoint is like walking out on the ice of a lake in winter. The further you go from the shore the more likely you are to be walking on thin ice and the more certain that when you fall through you will die.

      Statisticians tend to be more comfortable with interpolation than extrapolation. If, say, we have found a positive relationship between weight and heart disease in men of normal frame, by investigating weights varying from eight to fourteen stone; and you present with a weight within those limits I will be quite comfortable in interpolating and saying you have a greater or lesser chance of disease according as you are heavier or lighter. But if you present with a weight of twenty stone I should not extrapolate and say you are doomed to a quick death!

  14. January 23, 2021 7:05 pm

    Climate models cannot explain this warming if human-caused increases in greenhouse gases are not included as input to the model simulation.

    The logic being: the models can’t be wrong, so it must be those wicked humans. Genius 🙄

  15. Micky R permalink
    January 24, 2021 10:08 am

    Some excellent explanations on this thread re: inaccuracy of models, thanks.

Comments are closed.

%d bloggers like this: