Climate models

I came across this video (possibly from a tweet by Chris Colose) that describes climate models. I thought it was quite good, but I’m not a climate modeller so maybe others would disagree. The one thing I thought it did quite well was that it was careful to refer to what climate models do (wrt the future at least) as projections. You’ll often hear people claim that climate models have failed because observations have not matched model results over the last decade or so. Some immediate problems with this claim are that climate models are not actually optimised for decadal predictions/projections and what you’re often shown are ensemble averages, which tend to smooth out short-term variability.

A more fundamental point, though, is that when climate models are used to consider the future, they’re actually making projections, not predictions. One has to make assumptions about our future emissions (greenhouse gases, aerosols, …), solar variability, and other – as yet – unknown factors. Therefore, climate models are really telling us something about what might happen in the future if we follow various different pathways. Formally, therefore, if one wants to claim that a climate model has failed, one has to also know whether or not the pathway we actually followed was the same (or similar to) what was assumed in the model. Having said that, as Tom Curtis points out in this comment there is some indication that the model trends are different to the observed trends, suggesting that climate models may be too sensitive (I haven’t had much chance to look at this myself, so would quite like to know more about the significance of this). However, simply pointing out that the observations don’t match the model ensemble for the last decade or so, is not really sufficient to indicate that climate models have failed.

The other thing the video does quite well is talk about how climate models can be used to consider past changes. This can be used to try and understand what caused past climate changes, and can – presumably – also be used to see if these models can actually explain past changes. It also allows one to test for attribution. In the case of 20th century warming, models without anthropogenic forcings cannot explain the surface warming (both globally and regionally) after 1950. Models with anthropogenic forcings can. That’s one reason why the IPCC is very confident that most of the warming since 1950 has been anthropogenic. Anyway, I’ve said enough. I recommend watching the video yourself.

Advertisements
This entry was posted in Climate change, Climate sensitivity, ENSO, Global warming, IPCC, Science and tagged , , , , , . Bookmark the permalink.

42 Responses to Climate models

  1. Rachel says:

    Great video. Climate models are much more complicated than I thought. It’s hard to see how people can continue to argue that natural causes are to blame for the rise in temperatures. The video makes the case for human-caused warming very well I thought.

  2. Joshua says:

    Anyone –

    Having said that, as Tom Curtis points out in this comment there is some indication that the model trends are different to the observed trends, suggesting that climates may be too sensitive (I haven’t had much chance to look at this myself, so would quite like to know more about the significance of this).

    A distinction came up in the JN-G blog post related to this. When queried, JN-G said:

    she does not conclude that ECS is overestimated, just that the IPCC has lowered the lower end of its range and that there’s a sizable difference between observation-based and model-based estimates.

    I have the same question about Tom’s statement. When he said that the climates (typo) climate models may be too sensitive, is he saying that the IPCC has lowered the lower end of its range (not disputable), or is he referring to the speculation made recently by the IPCC that the discrepancy between observations and model projections may be due to over-sensitivity among other suggested explanations (such as increased OHC)?

  3. Joshua,

    is he referring to the speculation made recently by the IPCC that the discrepancy between observations and model projections may be due to over-sensitivity among other suggested explanations (such as increased OHC)?

    Yes, I think he is, and maybe Tom will clarify. I think the basic point is that the medium- to long-term warming trend is smaller than that suggested by most models. There are many explanations and quite a few implications. Some ideas that I’ve heard are :

    Aerosols – if we’ve underestimated the influence of aerosols, then models will run hotter than observations. This should have no influence on TCR/ECS because as we reduce aerosol emissions, the models will come back into line with the observed warming.

    Ocean response rate – we’ve underestimated how long (how much energy) it takes for the ocean to come into equilibrium with the rest of the climate system. This – I think – would mean the TCR is lower than the models estimate but may not influence the ECS, but it will take longer to get there.

    We’ve got some forcings/feedbacks wrong – this could influence both the TCR and ECS.

    Natural variability – it’s all just natural variability and after the next big El Niño models and observations will come back into line.

    I may be wrong about some of these and may be wrong about the implications, so if anyone else knows more, feel free to chip in.

  4. BBD says:

    Since policy never really goes away, it’s always worth bearing in mind that the sensitivity debate doesn’t include low enough values for TCR to mean we can just stop worrying and go to the pub. If only…

  5. BBD,

    sensitivity debate doesn’t include low enough values for TCR to mean we can just stop worrying and go to the pub.

    Indeed. If we’ve already seen 0.9oC of warming since 1880 and if Cowtan & Way are correct, the we could see (without any change in trend) a total warming of 1.3oC by 2150 (when we could have double CO2). Another big El Niño and we get more. Model estimates for the TCR are between 1.8 and 2 (for mean/median values I think) and so, realistically, even if they’re are running hot, it’s not going to be by a huge amount unless we start to see some cooling sometime soon (which I would think is rather unlikely).

  6. JCH says:

    If temp starts going up, which would be an observation, then observationally based estimates of TCS also have to go back up, right?

  7. BBD says:

    And then there’s paleoclimate…

    There’s just no way off the hook as far as I can see, although we can argue about tenths of a degree Celsius until Hell freezes over. Except it won’t, as ATTP points out above.

  8. izen says:

    The video is a tale of two halves. First its 2001-monoliths and super-complex equations that almost all the viewers don’t understand (including me!). The voiceover with the ‘wow science’ visuals states they can project the future climate and guide policy.
    The second part is a good overview of how models used to hindcast show the fingerprint of CO2 warming providing very strong evidence for attribution.

    The divergence between the modelled climates with and without anthropogenic influences were clear and the close correlation with observations explicit. The increasing uncertainty as finer resolution in time and region was also mentioned.

    Models are good tools for developing an understanding of how the system works. They are rather better at providing explanations than descriptions. Forgive a stubborn empiricist, but I think I prefer paleoclimate data to model projections for con straining future events.

    But consider if computational technology had been a few decades ahead in the 1960s. If the models we have now that show that fingerprint of A-CO2 and provide the evidence for attribution had projected then….
    How much doubt would we cast of climate modelling now.

    I suspect a part of the excited debate over model-observational discrepancies, that may be quite technical issues of timing and the resolution of aggregate model runs, is part of the doubt cast on the very existence of TCR, ECS etc…
    The rejectionist meme of complete dismissal of models as tunable simulators that can produce any alarmist prediction our lizard overlords require is aided by an expectation that modelling can provide a detailed description of the future. Casting doubt on the foundational physics that underlies the models when there is a discrepancy between models and observations is a tactic.

    The better the models can explain past climates, with accurate projections of what evidence might be found to support those paleoclimate explanations, the more certain we can be that future projections are reliable.

    But if they are as good about the next fifty as they are about the last, disputes over a 20% extension of the possible range of sensitivity downward might be slightly trivial.

  9. I think its an OK video, as long as you accept a narrow use of “climate models”. Its missing entirely the “use climate models to understand how the climate works”. Its only interested, really, in prediction/projection/attribution. Which is indeed part of their use; but its also a trap virtually all public talk of modelling falls into.

    It doesn’t really tell you what fingerprints are; probably, people will think the splitting-wiggly-lines-into-regions is fingerprints, which it isn’t.

    I think it could try harder to distinguish the “dynamical core” elements (the fluid dynamics bits) from the many parameterisations.

  10. izen says:

    John N-G had a good post on the complexity of modeling –

    http://blog.chron.com/climateabyss/2013/06/looking-under-the-hood-of-a-climate-model/

    The comments touch on model accuracy, while the latitude he may grant to some views may be ambiguous, there is clearly an anti-science line which if crossed, as with his logical post on Dr Judith Curry, elicits a response that can be quite … pithy.

  11. William,
    You’re probably right that it could have done a better job of certain aspects. The question, though, would then be whether or not it will still do as good a job of getting the basic message across. As far as I can tell, if you really want to start an argument amongst scientists, start a discussion about science communication 🙂

  12. Tom Curtis says:

    Joshua, I base my comments on a direct comparison of model outputs to Foster and Rahmstorf adjusted trends. It is possible that F&R got that wrong, but I have seen no convincing evidence of that. My conclusion, however, was not that model climate sensitivities are too low, but that there is growing evidence that they are too low. For how weak a condition that is, see my comments to Dana here. In fact, I tend to place more weight on paleo estimates of climate sensitivity than either model or recent estimates, and paleo estimates (on balance) suggest the models are getting it about right. However, if you look at the evidence available as at AR4 and that available now, there has definitely been an increase in the probability of lower (2-3 C per doubling) relative to the higher (3-4 C per doubling) estimates.

  13. Curious George says:

    I was very impressed with some parts of the video: In the modeled temperature I could see the Gulf Stream and the different climates of India and Tibet. Congratulations to the modelers.

    There were impressive-looking partial differential equations that the models solve. I only noticed a diffusion equation – maybe. The equations apparently describe the physics of the atmosphere, oceans, and land. Yet the video insists that the climate modeling is very different from weather forecasting. Which equations would be different for weather?

    Instead of freezing video frames containing equations , I propose that we discuss
    http://www.cesm.ucar.edu/models/cesm1.0/cam/docs/description/cam5_desc.pdf
    but please feel free to suggest another reputable source.

    With all those impressive equations, UCAR neglects a temperature dependence of a specific heat of water vaporization (and condensation) and does not know the effect of this approximation. As climate change is all about energy flow (and a change of 5K from the base of 385K is only 1.3%), an error of 2.5% in energy transfer between tropical seas and the atmosphere due to evaporation can not be dismissed out of hand.

    Finally, I have a big problem with the “models with human influence versus models without human influence”, showing the latter always below observed temperature, and the former agreeing with observations. Where can I find that in IPCC reports? All IPCC graphs that I remember show models projected temperature a way above observations.

  14. Curious George says:

    Sorry for a typo – a change of 5 degrees K our of a base of 285 K is 1.75%.

  15. Curious,

    Finally, I have a big problem with the “models with human influence versus models without human influence”, showing the latter always below observed temperature, and the former agreeing with observations. Where can I find that in IPCC reports? All IPCC graphs that I remember show models projected temperature a way above observations.

    It’s certainly in the SPM. Figure SPM.6 is – I think – one of those shown in the video.

  16. Brigitte Nerlich has a new post on Making Science Public about how images and short video cliips can influence discussions on comment threads. Chris, who sometimes comments here, has an interesting comment about how some dubious arguments can have similar themese across many topics and hence what satirists use for one topic may be relevant/appropriate for another. His example was the Monty Python “what have the romans done for us” sketch, which he associates with what might happen in a discussion with someone about how good climate models have been. Since it does seem like an apt comparison, it’s very funny, and because it’s relevant to this post, I thought I’d add it here.

  17. Curious,

    error of 2.5% in energy transfer between tropical seas and the atmosphere due to evaporation can not be dismissed out of hand.

    But I don’t think what you’ve said here is quite correct. If I understand it correctly, there is potentially a 2.5% error in the latent heat values used in these regions. Therefore, whether this has a significant effect or not depends on how much of the energy is transported via convection. If it’s small (as I think it is) then a 2.5% error in something small is not very important. If it’s significant, then it may be important, but I’m not aware that convection plays a significant role in energy transport through the atmosphere.

  18. > Which equations would be different for weather?

    None. What tends to be different is the resolution – forecast models are run for synthetic days not decades (although they have to be run in less real time), so can afford higher spatial resolution. But the real difference is that forecast models are initialised from a state that represents the real atmosphere at that point, and so the forward evolution represents the real weather.

  19. Curious,

    Yet the video insists that the climate modeling is very different from weather forecasting

    Just to follow on from what William said, the video actually said “the objective” is different, not that the model equations are different.

  20. Curious George says:

    Evaporation and convection .. we are both guessing. But that’s the only way hurricanes are powered; do you know of any other way?

  21. Curious,
    As I think I said in one of my earlier response, the wrong latent heat value could influence the amount of water vapour in the atmosphere and could also, as you point out, influence how well climate models model cyclone evolution. However, I don’t think that convection of energy through the atmosphere is particularly significant (otherwise, why does σT4 do such a good job of representing the energy loss from the planet). So, from an overall warming perspective, I can’t see how a 2.5% error in one small aspect of the models can have a big effect. Even when it comes to cyclones, if we’re interested in how they change in a warming world, I also can’t see how such an error can have much of an effect, if we’re interested in trends.

  22. Curious George says:

    Why does σT4 do such a good job of representing the energy loss from the planet? Because that’s the only way to lose energy into space. Actually, the problem with greenhouse gases is that they cause a deviation from a blackbody radiation, so strictly technically your quote of Stefan-Boltzmann law is out of context.

    I agree that we are interested in trends – but not in trends of an unknown. It reminds me of Himalayan climbers watching their mountain for a week and recording avalanches to map the safest route. With climate we surely want a better understanding of underlying causes.

    The video actually said “the objective” is different, not that the model equations are different. That sentence does not make any sense to me. Why don’t we use the model to produce a weather forecast for 100 days, instead of a climate forecast for 100 years?

  23. Curious,
    Yes, I know my blackbody quote was simplistic, but I still think BB radiation is the dominant energy transport mechanism from the surface. Happy to be corrected if someone knows better.

    Why don’t we use the model to produce a weather forecast for 100 days, instead of a climate forecast for 100 years?

    I think this is easy to answer. There a basic condition in numerical modelling called the Courant condition. It’s basically Δt < Δxv. If you want to do weather forecasting then Δx is small (i.e., you want to forecast locally) and so your numerical timestep (Δt) has to be small. For climate modelling you can use a coarser grid and so can use a larger timestep. We don’t have the computing power to do climate modelling at the same resolution as weather forecasting. We could do something inbetween (100 days) but again you’ve have to sacrifice some spatial resolution in order to run your simulation in some reasonable time.

  24. Curious,
    I should also have added that weather is thought to be chaotic and so making predictions beyong 5-10 days is pretty tricky even if your model assumptions and equations are perfect. Long-term climate is thought to not be chaotic and so making projections for the coming decades is thought to be possible. Making accurate predictions for the next 3 months is, however, pretty difficult, even if you could run a simulation with the required resolution.

  25. BBD says:

    Clearly time for the Hansen-is-a-model-sceptic quote again. Apologies to those already familiar with this.

    Yes, that’s right: James Hansen is sceptical about climate models:

    TH: A lot of these metrics that we develop come from computer models. How should people treat the kind of info that comes from computer climate models?

    Hansen: I think you would have to treat it with a great deal of skepticism. Because if computer models were in fact the principal basis for our concern, then you have to admit that there are still substantial uncertainties as to whether we have all the physics in there, and how accurate we have it. But, in fact, that’s not the principal basis for our concern. It’s the Earth’s history-how the Earth responded in the past to changes in boundary conditions, such as atmospheric composition. Climate models are helpful in interpreting that data, but they’re not the primary source of our understanding.

    TH: Do you think that gets misinterpreted in the media?

    Hansen: Oh, yeah, that’s intentional. The contrarians, the deniers who prefer to continue business as usual, easily recognize that the computer models are our weak point. So they jump all over them and they try to make the people, the public, believe that that’s the source of our knowledge. But, in fact, it’s supplementary. It’s not the basic source of knowledge. We know, for example, from looking at the Earth’s history, that the last time the planet was two degrees Celsius warmer, sea level was 25 meters higher.

    And we have a lot of different examples in the Earth’s history of how climate has changed as the atmospheric composition has changed. So it’s misleading to claim that the climate models are the primary basis of understanding.

    Yes, Curious, this was for you.

  26. Curious George says:

    BBD – thank you. I really am simply curious. If computer models are “our weak point”, say so, and don’t rely on them almost to the exclusion of everything else. It offends me. You may be right, and using weak arguments may make you look wrong. BTW, it is nice to know that Dr. Hansen echoes Dr. Trenberth’s communication to me.

  27. Curious George says:

    “Long-term climate is thought to not be chaotic.” A statistical average of a chaotic system is thought not to be chaotic. I’ll have to revisit my thought processes. Never too late to learn.

  28. Curious,
    Just so that it’s clear, I agree with what BBD was trying to illustrate. The fundamentals of AGW don’t really on climate models. Also, climate sensitivity (ECS for example) also does not rely only on climate models. There’s paleo-climatology estimates and energy budget estimates (which do rely on model forcing estimates but are largely determined by observations). So, climate models provide an extra estimate for climate sensitivity and provide information about how warming may influence our future climate. But to suggest that evidence for climate change relies mostly on climate models is wrong. I don’t know if this is what you’re implying, but there’s much more evidence for global warming/climate change than climate models alone.

  29. Curious,

    A statistical average of a chaotic system is thought not to be chaotic.

    That may well be a better way to put it, but I do think most would argue that climate trends are likely not chaotic.

  30. Curious George says:

    Our views are (I hope) slowly converging. There is plenty of evidence of global warming/climate change. We no longer live in an ice age. Thirteen thousand years ago the ocean level was 130 m lower than today; the Black Sea was a fresh (how do they know?) water lake. I remembered easily these two multiples of 13; today we probably have better data – this shows you that I am an Old Timer. As such, I also vividly remember the scare of global cooling. Please call me a skeptic.

  31. Curious,
    Maybe we are converging, but

    As such, I also vividly remember the scare of global cooling.

    do you agree that if there ever was a scare of global cooling it was in the media only, and not in the scientific literature. It’s my understanding that even in the 1970s, the number of papers that were suggesting global warming exceeded the number suggesting global cooling.

    Also, there is a difference between global warming that has resulted in us slowly moving out of an ice age and anthropogenic global warming that – the evidence suggests – is rapidly (on past climate timescales) moving us into a climate regime that is unprecedented in the Holocene.

  32. BBD says:

    Curious George

    BBD – thank you. I really am simply curious. If computer models are “our weak point”, say so, and don’t rely on them almost to the exclusion of everything else. It offends me.

    As Hansen and ATTP point out, this is not the case at all.

    Hansen & Sato (2012) provides an empirical estimate of climate sensitivity which I recommend to you. They find ECS/2xCO2 to be 3C +/-1C, and state:

    This empirical climate sensitivity incorporates all fast response feedbacks in the real-world climate system, including changes of water vapor, clouds, aerosols, aerosol effects on clouds, and sea ice. In contrast to climate models, which can only approximate the physical processes and may exclude important processes, the empirical result includes all processes that exist in the real world – and the physics is exact.

    Several people have made the point that the uncertainty does not preclude a policy response. So how much weight should be attached to discussion of the uncertainty is moot considering the size of the elephant over there.

  33. jsam says:

    On a different side of climate modelling, if Lomborg and Tol say we’re in trouble then we’re in big trouble.

    http://www.theguardian.com/environment/climate-consensus-97-per-cent/2014/jan/24/more-global-warming-worse-economy

  34. Joshua says:

    As such, I also vividly remember the scare of global cooling.

    Oh, brother.

    Curious, do you not realize how unskeptical that argument is?

  35. jsam says:

    Why don’t so called “sceptics” show some scepticism and research their assertions before posting them? Curious by name, perhaps, incurious by nature.

    https://www.skepticalscience.com/What-1970s-science-said-about-global-cooling.html

  36. jsam,

    On a different side of climate modelling, if Lomborg and Tol say we’re in trouble then we’re in big trouble.

    Personally, I would argue that whether Lomborg or Tol say we’re in trouble or not has virtually no bearing on reality.

    I read the Guardian article, but haven’t read the report so don’t really know if how the Guardian article interpreted the data is the same as how the report interpreted it.

  37. Curious George says:

    BBD – (Hansen and Sato, 2012): the [empirical] physics is exact. Thanks a lot. I’ll need some time to study it.

  38. John Mashey says:

    JSAM:
    Skeptics were people like Carl Sagan or Martin Gardner, and most scientists, most if the time.

    Most “climate sceptics” / dismissives (as per Yale/GMU SIX Americas)may (or may not be) skeptics in other areas, but in climate they are *pseudoskeptics”, claiming the venerable skeptic label, while employing powerful Morton’s Demons, which allow the silliest claims that support the worldview, including bizarre paranoia/conspiracy ideation that confirms Lewandowsky, while rejecting even simple facts that challenge them, and adhominem”ing people who bring such facts.

  39. Reich.Eschhaus says:

    Poor blog post. Has less than 40 comments. Must be rubbish! 😛

  40. Reich, do you think I judge my blog by the number of views and the number of comments? Who do you think I am? Anthony Watts? 🙂

  41. Paul S says:

    Curous George,

    ‘Finally, I have a big problem with the “models with human influence versus models without human influence”, showing the latter always below observed temperature, and the former agreeing with observations. Where can I find that in IPCC reports? All IPCC graphs that I remember show models projected temperature a way above observations.’

    Those kind of features can be strongly influenced by choice of baseline. In general observed warming trends lie at the upper end of the model ensemble for the first half of the 20th Century and the lower end for the past few decades.

    The graphs you’ve seen with models overshooting observations have probably used a 1961-1990 or even 1986-2005 baseline. These choices mean that anything which happened prior to 1961 or 1985 are ignored in terms of the match to recent temperatures, which provides greater opportunity for shorter term variability to influence the apparent match.

    The SPM graph referenced by ATTP uses a 1880-1919 baseline, which allows a longer term perspective on how models and observations have varied together, smoothing out shorter term influences and discrepencies over the course of the century.

  42. Reich, 41 now (well, 42 with this one) so post improving. Of course, if I correct for self-citations then …..

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