Great Moments in "Science"

You know that relative of yours, who last Thanksgiving called you anti-science because you had not fully bought into global warming alarm?

Well, it appears that the reason we keep getting called "anti-science" is because climate scientists have a really funny idea of what exactly "science" is.

Apparently, a number of folks have been trying for years to get articles published in peer reviewed journals comparing the IPCC temperature models to actual measurements, and in the process highlighting the divergence of the two.  And they keep getting rejected.

Now, the publisher of Environmental Research Letters has explained why.  Apparently, in climate science it is "an error" to attempt to compare computer temperature forecasts with the temperatures that actually occurred.  In fact, he says that trying to do so "is harmful as it opens the door for oversimplified claims of 'errors' and worse from the climate sceptics media side".  Apparently, the purpose of scientific inquiry is to win media wars, and not necessarily to discover truth.

Here is something everyone in climate should remember:  The output of models merely represents a hypothesis.  When we have complicated hypotheses in complicated systems, and where such hypotheses may encompass many interrelated assumptions, computer models are an important tool for playing out, computationally, what results those hypotheses might translate to in the physical world.  It is no different than if Newton had had a computer and took his equation Gmm/R^2 and used the computer to project future orbits for the Earth and other planets (which he and others did, but by hand).   But these projections would have no value until they were checked against actual observations.  That is how we knew we liked Newton's models better than Ptolemy's -- because they checked out better against actual measurements.

But climate scientists are trying to create some kind of weird world where model results have some sort of independent reality, where in fact the model results should be trusted over measurements when the two diverge.  If this is science -- which it is not -- but if it were, then I would be anti-science.


  1. mesocyclone:

    A climate modeling acquaintance of mine proposes a somewhat plausible, and certainly convenient answer to the climate pause: the models do show periods of temperature stability. They just don't show them for the climate pause period.

    I find that too convenient, but it is consistent with model logic: that the models may be right in the long term (as the variations are integrated over time) but rarely right in the short term.

    It's handy that this is not testable, but it is a sort of response to the climate pause criticism, and is in line with the ERL publisher's comments.

  2. mesaeconoguy:

    I made $20,000,000 this week trading.

    Remember, it is important not to compare actual valuations which may have occurred with my model portfolio gain/loss (generated using hypothetical asset values).

    The actual valuations are misleading, and can lead to oversimplified interpretations of my financial condition.

  3. August:

    They really don't listen. I thought Climategate was finally going to end it, but these guys just decide on their narrative and stick with it, like Benghazi was just about a movie. Benghazi is a telling one, because one that may have been our ambassador, but it was one of their guys- solidly in the leftist camp. They should care, because they could just as well have gotten sacrificed like that, but they just stick their finger in their ears.

  4. Arrian:

    What is considered "long term" anyway? I can see a year or three being short term, but when the slope is off trend for a decade, isn't that getting into a long enough term to ask "what are we missing?"

  5. mesocyclone:

    No, it isn't.

    I'm not saying that this view is correct, btw. I'm just reporting it.

    He applies the same logic to model chaos. GCM's suffer from chaos. They are basically weather forecast models, and anyone who has ever used them know that they have virtually no skill after 5-10 days. But, what they do forecast may be wrong, but it is physically possible (this is the math of chaos at work). Thus if you run the model long enough, and no other sources of error sneak in (a questionable idea), the long term average will be right in some sense.

    This is analogous to a common weather modeling practice: ensembles. Because the initial conditions for the model are not precisely know, the model is initialized with a best guess. With ensembles, you run the model a bunch of times, using different guesses, all of which are within the error bars of the actual data. You may also run different versions of the model or different models. Then you look at the entire group of models, and it turns out that doing this extends their skill range farther into the future than a single run.

    The link below gives an example from an operational run today. Each different color line is the output of a different model run. This shows a low pressure trough over the western US - at least, most runs do. It also nicely demonstrates chaos.

  6. alanstorm:

    "that the models may be right in the long term (as the variations are integrated over time) but rarely right in the short term."

    I've heard that before - it's the same spiel used to keep the useful idiots idiotic, proclamations about "The inevitability of World Socialism" despite minor setbacks like losing the cold War.

  7. Pinebluff:

    Follow the money. The secret back door output of the models is grant requests. Without the government funding these clowns might have to get real jobs.

  8. mesocyclone:

    Yeah, except here we are talking about science, not sociology. Modeling is complex. Ensembles do help. If climate models can legitimately be looked at as serial ensembles, that's interesting.

  9. marque2:

    17 years and 9 months. It has actually been cooling on average for that last 12 years and that is with NOAA adjusted numbers. They can no longer adjust the numbers up fast enough to hide the decline.

  10. marque2:

    Weather reports are way different from CO2 climate models. They are also designed to measure different things and can't really be compared.

  11. Chris:

    This is the argument I've used. No one can model the financial system with any degree of certainty and there are orders of magnitude fewer data points.
    No one in their right mind would give any climate guesser a penny if they were shown climate models and told they were market models.

  12. mesocyclone:

    Climate models almost all use the same technology as weather models, and in one major case, the same Fortran source code. The main difference is that they are run far into the future. The modification to add in the effects of changing CO2 levels are very minor.

    There's a reason they tend to be the same: they are modeling the same physical system: the atmosphere, and more recently the ocean. The primary techniques for doing this involve finite element modeling - you cut time into discrete slices, and cute the geometry into three dimensional boxes. You then add in "parameterization" to inform the model about the expected results of sub-grid processes. Then you start cranking the model. There are variations (such as Fourier modeling of some components) but generally, you have to use the same technology for both.

    You can get an introduction into both weather and climate modeling here:

  13. mesaeconoguy:

    Actually worse than that Chris, if any investment manager used similar tactics as the warmists, FINRA and SEC (and probably others) would sanction them and possibly prosecute them.

    I say possibly because we now live in a country with selective application of law.

  14. mikehaseler:

    That's the difference between skeptic science and consensus science facts versus winning climate wars

  15. mesaeconoguy:


    Do they still use the punch cards?

  16. marque2:

    Believe me, they published the code for one of the models during climate gate. The code was very simple to the point of being tivial and was not concerned with day to day weather.

    With day to day weather, they quickly calculate what the weather will be in the next ~= 15 minutes, which can be done fairly accurately and then use that data to calculate the next ~= 15 minutes, detailed info about clouds and wind speed, air pressures, etc are used in the model. The weather model is considered to be accurate to 50% after about 15 days of projection. The long term CO2 increases are not even relevant. And running the weather models longer just produces junk and it is know to be junk.

    But believe what you want, Yeah, those climate scientists are producing very fine long term models exactly the same way the produce our 10 day weather forecasts. I bow and pray to the great Gaea.

  17. marque2:

    This is the reason they don't want the CO2 long term prediction software to escape in the wrong hands - why they don't allow peer review of the code by outside experts. It is just trash code. I found a Watts Up that article that goes over the code from the one model that escaped.

  18. marque2:

    Probably. You would be surprised what old stuff you find hanging around offices. Sometimes they need the old equipment just in case some old data or document gets discovered and needs analysis. Or 29 years after software is written a bug is found that needs repair (just before the 30 year lifecycle ends). Also it is frequently cheaper to just continue to upgrade software on the older equipment than it is to actually move the software to newer equipment change compiler, language, etc, because safety critical software would need to be re-certified and that gets outrageously expensive.

  19. marque2:

    Correction, modelling should be complex. All the modelers are told that the reason temps go up is because of CO2, so they don't model in the other effects, like sun cycles causing differences in cosmic ray bombardment (increasing cloud cover). So they do some minor temperature stuff for the world, throw in X amount of heating due to CO2, and wherever the model doesn't work, they sprinkle in more Aerosols, because aerosols cool. So 75 - 78 doesn't seem to fit, Oh there must have been more aerosols. l think that is the current excuse as well.

  20. marque2:

    Hey may I use your avatar? :-)

  21. mesaeconoguy:

    Science, like Nature

    Must also be tamed

    With a view towards its preservation

    Given the same

    State of integrity

    It will surely serve us well

  22. mesocyclone:

    I will respond to both of your comments here.

    I am not a climate alarmist, but I do try to understand how they arrive at their results.

    You should learn the difference between paleoclimatology (where the models are simple and *only* analyze past climates) and climate modeling, which attempts to model future climate. You have addressed only the former, when it was obvious that I was writing about the latter - at least to anyone who pays attention to this stuff.

    You are beclowning yourself on this topic because you haven't a clue what you are talking about. I gave you a link to a place where you can actually learn.

  23. mesocyclone:

    A lot of science is done in FORTRAN. It's ancient and clunky, but it's used. I stopped using it 40 years ago, but then I mostly don't do scientific computing.

  24. mesocyclone:

    Correction: climate modeling is complex.

    You are confusing statistic models of past climatic data with "climate models." They are not even close to the same thing. The former are relatively simple. The latter *start* with already hugely complex weather models (or weather model techniques) and then build in even more complexity. If they were simple, as you say, then why do they have a whole bunch of supercomputers - the most powerful that NOAA has?

    One gripe I have with the climatetistas is that they are getting all the high powered computers, leaving the weather service, which actually does forecast weather for hundreds of millions, with inferior hardware and resulting poor models.

  25. Matthew Slyfield:

    " If they were simple, as you say, then why do they have a whole bunch of supercomputers - the most powerful that NOAA has?"

    1. Status
    2. Money
    3. Because if anyone realized you could run their models on a 15 year old PC they would be laughed out of their jobs.

  26. mesaeconoguy:

    Well, at least they don’t use BASIC…

  27. mesaeconoguy:

    Whoa, mesocycloneguy, I liked your link, and watched some of it, but it was below my scientific comprehension resolution.

    I want the complex multivariable calculus version, not some AGW punchcard mechanic version.


  28. mesaeconoguy:

    Mesocyclone is correct: multivariate modeling is extremely complex, and is very hard to do.

    Most complex dynamic systems, such as atmospheric, financial and commodities market, and other similar domains cannot be reliably forecasted, even with the best computational models.

    I used to build several, then got smart.

  29. HenryBowman419:

    Sorry, I don't think anyone would pay them for their supposed "expertise". I've worked with plenty of these clowns: they have no future outside of government funding, and they know it. That's why they become desperate very quickly.

  30. mesocyclone:

    Heh. I'm sure that's available too. But I'm not deep enough into the field to want spend a lot of time on the mathematical details Besides, my vector calculus is pretty rusty.

  31. mesocyclone:

    I don't know.. in the case of Climategate. For all i know, a bunch of their stuff is in Excel spreadsheet formulas. The next step *up* from that is VB.

  32. marque2:

    One this about the alarmists - they have nature on their side. The last two interglacials reached temperatures 2degC higher than current temps and the ocean levels were 24 feet higher. This was right before a precipitous drop in temps.

    It might take 3-400 years but these numbers will be reached no matter what we do and they will all claim victory - until ...

  33. mesaeconoguy:

    Mine too, cease fire.

    Tis a curious thing, which brings men to speculation.

  34. mesaeconoguy:

    I think they were at least C++

    Into Excel, csv format.

    Anyway, all current climate models have failed.

  35. Zachriel:

    Coyote Blog: Apparently, in climate science it is "an error" to attempt to compare computer temperature forecasts with the temperatures that actually occurred.

    It's says the paper was comparing apples with pears. Energy balance equations don't directly predict temperature. Another problem was that the paper didn't attempt to explain the discrepancies. The reviewers also pointed to technical errors. Finally, the "overall innovation of the manuscript is very low", basically just a rehash.

  36. marque2:

    Never said it wasn't complex - if they actually really tried to do it. Most of them have - if Co2 raise temp, if past years are too high add aerosol - that isnt very complex at all. And for all the BS about how they do complex modelling, we don't know - because they keep the models AMD software as closely guarded state secrets - so no real reviews are done to figure if they actually do anything.

    The only model we have is from climate gate and it is about as primitive as you can imagine.

  37. marque2:

    I think you are confusing modeling with science.

  38. marque2:

    Interesting fact - typical top end cell phone has much more power than the original Cray 2 supercomputer. At the time that was the fastest in the world. Phone has 4x the megaflops.

    This is very useful and necessary to the development of car videos.

  39. John Warstorm:

    Well great moments in Science for me is new discovery that is very beneficial to mankind...

  40. morganovich:

    what a whitewash.

    so, if they do not predict temperature, then how can they be used to predict warming and why do we even care? you have just cornered yourself here. either the models can predict temperature, and therefore can be compared to observed temperatures, or they cannot predict temperature, in which case, those using them have no business making claims about warming

    further, as ar4 and 5 did make very specific clams about temperatures, your "energy balance" claim is just obfuscation.

    this graph showing temperature change estimates come straight out of ar5.

    claiming that such predictions should not be compared to observed temperatures is simply absurd.

  41. Zachriel:

    morganovich: if they do not predict temperature, then how can they be used to predict warming and why do we even care?

    Because while the energy of the Earth's hydrosphere, cryosphere, and atmosphere increases, different parts of the system absorb heat at different and varying rates.

    morganovich: either the models can predict temperature, and therefore can be compared to observed temperatures, or they cannot predict temperature, in which case, those using them have no business making claims about warming

    Increasing heat content can be largely predicted, but how that heat is distributed as the Earth warms is chaotic.

  42. Zachriel:

    morganovich: this graph showing temperature change estimates come straight out of ar5.

    The observed temperatures are within the error bars.

  43. marque2:

    You know why your theory that short term and long term forecasts are the same. The reason is, that marginal changes in the elemental component of air have little consequence on the 10 day forecast, it probably isn't even a component of the forecast. Even things like aerosols, or long term changes in Sun output do not matter, because they are just tracing how weather conditions flow.

    You need an entirely different model to include the effects of trace gasses in the environment, Cosmic Rays, sun output, etc. (Of course the models are commissioned with the insistence that CO2 is the driver, so the other factors I mentioned are largely ignored. ) And of course, unlike the 10 day forecast, the fact that a cloud is, or will be traveling over Sacramento is irrelevant, it is average cloud cover and average temperatures over certain regions that they are concerned about.

    Anyway, it is incredibly naive to think that your day to day weather models and long term models are the same, or even similar.

  44. mesocyclone:

    @marque2:disqus , "it is incredibly naive to think that your day to day weather models and long term models could be the same, or even similar."

    The models are *not* commissioned with CO2 as the driver. The models are commissioned to attempt to simulate the system. CO2 is tossed in as a minor tweak. Cosmic rays are not put in because their contribution is not understood.

    Changes in CO2 concentration and sun output are easily put into GCM's. They are one tiny component of an enormously complex system modeling the atmosphere and, these days, the oceans. To say that they could not be similar to weather modeling simply demonstrates a complete lack of understanding of models.

    I sure hope you don't post to the more public forums on this subject! Your lack of knowledge, and your certainty of your opinions hurts the credibility of the skeptic cause. As a climate skeptic, I would rather that uninformed skeptics STFU.

  45. marque2:

    Sorry dude - I believe you are ignorant or gullible. It is sad that I as a climate skeptic have to read such nonsense from self proclaimed experts who claim to know.

    Of course long term forecasts that pretend to show CO2 is forcing the climate do not use CO2 except in a minor way. Major eye rolls.

  46. mesocyclone:

    The ocean/atmospheric system is very big and very complex. CO2 is a minor input. But... the forecast warming is minor, too - 2-4 degrees out of 270 or so. So save your eye rolls, because the impact is minor, and the impact on modeling is a lot more minor, but that minor effect is what everyone is talking about.

  47. Michael:

    Why should we even care about the warming and the "rise of the oceans". I remember that on every school trip I went (Israel) the guide told us that once upon a time this desert was underwater.

    They say the Poles will be warmer but the Equator temp. will be roughly the same.

    A few points:
    1) Melting of the North Pole would clear new sea routes and resources underwater,
    thus boosting the economy of most northern countries.
    2) Withdrawal of the permafrost line would clear vast areas of fertile land to agriculture and living
    (Canada, Russia, Scandinavia)
    3) Higher CO2 levels help plants grow.

  48. FelineCannonball:

    Models are only good for what they are meant to model, and climate models were never intended or capable of reproducing annual or decadal oscillations accurately. The best global climate models may accurately reflect some of this complexity, but because of extreme sensitivity to starting conditions, purturbations, or parameters will mean the scale of these oscillations but not their timing will be reflected in the output. Run it 10000 times and maybe you'll find a few that do well through the last 20 years. The right way to look at these models is to look at the envelope that would contain 99% of the model runs and remember that short term variation is chaotic.

    Dripping faucets are chaotic, Weather is chaotic, El Nino is chaotic, decadal oscillations are chaotic, long term climate in the thousand year time scale is forced. Gleick's Chaos might be a good read.

  49. FelineCannonball:

    I get the annoyed at the folks who want to compare the model average or a single model run to the real world. Imagine how useless that would be with hurricane tracks. If you averaged the model runs you'd predict a diffuse strong gale along a large stretch of coast. If you used a single model run you'd confidently predict a hurricane hit on a particular spot on the coast, ignoring the chaotic nature and instability of the system.