Climate Models

How well have climate models forecast global temperature?

I took the data for global average temperature from climate model simulations in the CMIP5 archive; those are computer models used in the latest IPCC report. I used only those models with the “RCP4.5” emissions scenario (a middle-of-the-road choice). I then aligned them all so their average value was zero during the 1961-1990 “baseline” period. Finally, I calculated yearly averages for each of the 108 models included.

That enables me to compute the “multi-model mean,” the average of all the models at each moment of time. Also at each moment of time, I computed the standard deviation of the model values and recorded the highest and lowest model values (which can be different models at different times).

Now I can graph the multi-model mean over time as a thick red line, together with a yellow outermost envelope showing the range from highest to lowest, a tan-colored middle range the limits of the 2-sigma range (about 95% of the models) and a pink band the 1-sigma range (about 2/3 of the models).

And I can also plot actual observed global temperature from NASA (yearly averages using the same 1961-1990 baseline) as a black line:

Recent values are genuine predictions, in the sense that the observed values of global temperature weren’t known when the models were run (it can take a long time to run these models, even on a supercomputer).

Clearly actual temperature has followed the model results closely, staying within the 1-sigma range most of the time. The latest value (2019 year-to-date) is right in the bull’s eye.

Thanks to all those who generously contributed to the blog. If you’d like to help, please visit the donation link below.

This blog is made possible by readers like you; join others by donating at My Wee Dragon.


17 responses to “Climate Models

  1. Hi Tamino,
    Is this a like-for-like comparison with SAT+SST for both or are you comparing SATs to SAT+SST (e.g. Cowtan et al 2015)?

    [Response: It’s comparing SAT (from models) to SAT+SST (obsesrvations). A proper comparison (a al Cowtan & Way) makes the most recent observations even higher relative to the models.]

  2. How much should we worry about the climate feedbacks missing from the CMIP5 models, feedbacks that may be more powerful in the future. From an answer I received from the UK Department of Energy and Climate Change (just before it was abolished):

    In answer to your specific questions:
    1. Am I correct in thinking that some of these feedbacks were not used in the models that calculated the “remaining carbon budgets” – as used in
    the IPCC AR5?

    That’s correct, the models used vary in what they include, and some feedbacks are absent as the understanding and modelling of these is not yet
    advanced enough to include. From those you raise, this applies to melting permafrost emissions, forest fires and wetlands decomposition.

    2. Are there other missing feedbacks that should be considered?

    The feedbacks you mention are certainly important, although there are several other feedbacks that could be included but are currently too difficult to model. As knowledge and understanding advances, they will be added to the climate models

    Full text here

    • @geoffbeacon,

      As I understand it, modellers are working very hard at including ice sheet dynamics in the next round of climate models. (Source: Prof Syukuro Manabe, personal communication when he was visiting Harvard.) As you can imagine, that in itself is devilishly difficult.

      • @ecoquant
        Thanks, that’s good to hear.

        How is the remaining ice volume (RIV ) related to global mean surface temperature (GMST) ?

        This and the relation between GMST and ocean heat content (OHC) are interesting because of the role of the 2nd greenhouse gas, methane.

        The heating caused by methane affects RIV and OHC more than GMST because higher GMST causes heat to be radiated to space.

        Is this important?

        Any good papers on this?

      • @geoffbeacon,

        I have no papers specific to your questions.

        My understanding is that Arctic melt is driven by local ocean temperatures, elevated both because of overall oceanic warming and because of the mutually reenforcing albedo warming after melt and because of local atmospheric warming … Ice freezes, releases heat to air, causing storms (previously didn’t happen), and then heat gets sucked down when ice melts. Trouble is, ocean heat content increases all the time. Remember heat capacity of water.

        On the other question, remember why surface temperature is higher: Shift upwards of the tropospheric temperature profile maintained by the lapse rate.

      • Hi, Geoff. You say “The heating caused by methane affects RIV and OHC more than GMST because higher GMST causes heat to be radiated to space.” I don’t understand – instead of “more than GMST” do you mean “more than CO2”?

        PS – since ecoquant mentions Dr. Manabe, I’ll mention that today, by chance while renewing my subscription to the indespensible Science News, I came across an article from 50 years ago (11/15/69) that mentioned work by him and Dr. Richard Wetherald in an article discussing the cooling going on at that time.

        The discussion included this quote: “Work by Drs. Syukuro Manabe and Richard T. Wetherald of ESSA’s Geophysical Fluid Dynamics Laboratory has shown that at least a third of the warming from 1880 to 1940 can be attributed directly to the carbon dioxide buildup.”

        Referencing the cooling, this was mentioned: “But Dr. Reid A. Bryson of the University of Wisconsin points to studies indicating that the amount of dust and other particulate matter in the atmosphere has increased dramatically in recent decades, a change that could counteract the thermal effect of carbon dioxide buildup.”

        Article’s in their paywalled archive (but subscribing goes to the Society for Science and the Public).

        Frustrating that things known decades ago have been respun as controversial, when what we’re really waiting to find out is the final bill for our fossil fetish.

      • @b fagan

        Is this correct?

        1) A pulse of methane raises GMST for a decade or so. Higher GMST causes extra heat to be radiated to space so the excess temperature is dissipated . (The smaller effect of CO2 as a breakdown of methane ignored here.)

        2) The pulse warms the ocean but some of the heat is removed from the surface so is not so easily radiated into space. The effect on OHC lasts longer.

        3) The pulse causes warming that melts ice. Not much temperature increase due to this – so no extra radiation to space – but there is an ice-albedo feedback, producing much longer-term warming.

        Is there any good work discussing the balance of these effects?

        Are there other effects?.

      • Hi, Geoff. I think #1 is different (someone please correct me if so) – The methane is trapping heat and warms the surface and the atmosphere. Because methane traps different IR wavelengths than CO2 or water vapor, it’s blocking IR radiation to space, so the atmosphere would warm and expand, and the surface would warm also. Lapse rate means cooling the layer that eventually radiates to space. So like with any dose of GHG, it’s raising climate temperature as the system gets rid of more heat with less efficiency.

        I’m not aware of any papers that cover the balance atmosphere/ocean/cryosphere in one fell swoop, but the additional heat from methane would also amplify the reduction in surface albedo as more ice and snow melt away. The oceans would, as you mention, bank some of the heat, and the work of melting would be using some of the available energy that would otherwise heat air and surface, slowing the temperature increase an ice-free planet would experience with the same dose.

        I found a useful and brief discussion of methane vs. CO2 here:
        Mr. Byrne reminded me that the doubling effect is a bigger risk for methane since much less is needed to double it. He also linked to the good discussion in the IPCC reports (though IPCC broke all their old links, so you’d have to get to TAR WG1 chapter 6 by clicking through the new site).

        One thing to take a bit of comfort in is that, when the Berkeley Lab verified actual greenhouse effect changes tied directly to measured GHG concentrations for CO2 and methane, the methane study happened during a real climate pause – the years of slight decline in methane. That study showed the change in trend as concentrations again started rising.

        I don’t have access to Nature publications, but here are the press releases.

  3. Thanks Eli. To this interested amateur, your graph presents two interesting questions:

    first, the post-WWII plummet in global temperatures present in so many other graphs has disappeared. Better reconciliation of ocean temp records?

    second, what happened in 1970 (approx)? Just normal variation or is something else going on?


  4. Nice work!
    Much better than Ross McKitrick who (wilfully?) ignores the coverage bias in HadCRUT:

    Click to access model_obs_comp_nov_2019.pdf

    However, as mentioned above, the comparison could be even more apples-to-apples. Blended model ensembles (Cowtan 2015) are available at KNMI climate explorer. A quick look shows that Gistemp Jan-Oct 2019 is 0.02 C above the blended rcp4.5 multi-model mean for 2019

  5. I think it is important not to compare the global temperature anomaly calculated from the models with the surface temperature indexes. Because the surface temperature indexes are lacking coverage especially in regions with sea ice. This causes an underestimation of warming in the presence of decreasing sea ice, or an underestimation of cooling if the sea ice area increases.

    Therefore, because full coverage surface temperature measurements are not available, I would suggest to restrict the model output from global to the area where measurements are available, and using model SST where measurements based on SST anomalies. This should be a more realistic comparisons of measurements and model output.

    • Uli

      Sounds good!

      If I was not too lazy to process NetCDF data, I would do the job.
      But.. I prefer to stay on simple text files, and to enjoy free time.

  6. What do you think about this paper ?

    good science or bad science ?

    Click to access Zhu_et_al-2016-Nature_Climate_Change.pdf

  7. Hello Dr. Foster.

    The denialist Nicola Scafetta made some easily falsified claims in two of his recent papers on TSI (total solar irradiance). As an extra bonus, by Scafetta’s logic, falsifying his claims would undermine his TSI model. Scafetta’s basic claim is that global surface warming occurred from 1970 – 2000, while post-2000 temperature remained relatively stable, supporting his model of a 1970 – 2000 TSI increase and a post-2000 TSI decrease.

    Given your background analyzing surface temperature trends, I think you’d have no problem falsifying Scafetta’s claims. So below I’ve posted Scafetta’s claims, along with a Twitter thread in which I’ve debunked them. I won’t hold out much hope for Scafetta publishing an erratum or corrigendum to either of his papers, given his long track-record of uncorrected misrepresentations meant to manufacture false doubt regarding AGW.

    Section 4, page 12:
    “The global surface temperature of the Earth increased from 1970 to 2000 and remained nearly stable from 2000 and 2018. This pattern is not reproduced by CO2 AGW climate models but correlates with a TSI evolution with the trending characteristics of the ACRIM TSI composite as explained in Scafetta [6, 12, 27] and Willson [7].”

    Click to access 1214896.pdf

    Section 6, page 22:
    “Final evidence that TSI may have increased from 1980 to 2000 comes from Earth’s climate studies. Secular climate records correlate well with TSI curves such as the one depicted in Figure 13 and on longer ones covering the entire Holocene […]. In particular, the warming observed from 1970 to 2000, followed by a temperature standstill since 2000, is a good fit for a natural 60-year cycle prediction superimposed to other contributions”