Plausible emission scenarios

A paper by Roger Pielke Jr, Matthew Burgess and Justin Ritchie has been submitted that suggests that the most plausible 2005-2040 emission scenarios project less than 2.5oC of warming by 2100. It’s generated a bit of debate on social media, so I thought I might write a post with some thoughts.

What the paper does is compare emission growth rates in IPCC scenarios with observed growth rates over the period 2005-2020, or with a combination of observed growth rates and those suggested by a set of IEA scenarios that go out to 2040. Those that most closely matched were then assumed to be the most plausible when it comes to projecting emissions, and warming, to 2100. The most plausible suggest a change in forcing of about 3.4 Wm-2 by 2100, only a small fraction are consistent with 6 Wm-2 by 2100, and the even higher emission scenarios (RCP7.0 and RCP8.5) lie far outside the envelope of plausible scenarios.

However, as Ken Caldeira pointed out on Twitter, would you expect grow rates in the early part of a century to be a good predictor of century-scale emissions? Imagine going back to the early 20th century, developing a set of emission scenarios for the 20th century, and then basing their plausibility on how well they match early 20th century grow rates. How well do you think you would have done?

On the other hand, given that there are active efforts to limit how much we emit and that the Paris agreement has the aim of limiting global warming to well below 2oC (with aspirations to limit it to below 1.5oC) we may well now be in a position where our current trajectory is taking us towards lower emission pathways and, hence, lower levels of global warming. So, my general sense is that the basic result in the paper is not unreasonable.

However, this is a very complex socio-economic system. How much we emit in future depends quite strongly on what we do in future. We could be stupid and find some way to extract hydrocarbons from clathrates, making the higher emission pathways suddenly more likely, or we could be clever and implement effective negative emission technologies that suddenly make the very low emission pathways plausible. I think both are unlikely, but neither is impossible.

Also, this is not an independent system. If we suddenly believe that we’ve done enough to limit warming to 2.5oC and relax efforts to limit future emissions, the higher emission pathways become more plausible. On the other hand, if we believe that we should re-double our efforts to limit future emisisons, then the lower emission pathways become more plausible. I do think we should be cautious of making probabilistic claims about complex socio-economic systems.

Also, it appears that there are a number of factors that this paper ignores. For example, it seems to be considering only CO2 emissions from fossil fuels and industry. It doesn’t consider emissions due to land use change, or (as far as I’m aware) consider the emission of non-CO2 greenhouse gases, such as methane. Both of these are factors that can influence how much we would need to emit to reach a certain change in forcing, and hence level of global warming. Similarly, it doesn’t include carbon released through thawing of the permafrost, which would act as an additional emission source. There’s also some uncertainty when it comes to associating emissions with concentrations. We could end up at a higher forcing level even if we follow an emission pathway typically associated with a lower forcing level.

I’m not suggesting that this implies that the basic conclusion is wrong; I do think that our current trajectory is taking us towards something like ~2.5oC (± ~1oC) but I’m less convinced that the confidence in this is warranted. I do think the paper would have benefitted from a broader discussion of the various caveats and uncertainties. In fact, given that two of the authors have also published a paper suggesting that the misuse of scenarios is a research integrity issue, you might expect this to have been more explicit. Of course, if you’ve been involved in the public climate debate for a while, you probably wouldn’t.


For some reason, Roger decided to challenge myself and Richard Betts to a bet on the basis of his claim that “In 2040 global CO2 emissions from energy will be closer to levels projected by SSP1-2.6 than to SSP4-6.0”. This seemed a bit odd given that there seemed little reason to bet on something that we would rather didn’t happen, and have never suggested is necessarily all that likely to happen. However, it seems that this may have been an attempt to somehow test our arguments. If we’re suggesting that something close to SSP4-6.0 is plausible, but won’t take a bet on this actually materialising, then that supposedly says something about the plausibility. One should probably, though, convolve this with the probability of someone betting on an outcome that they neither want to happen, nor think is all that likely to happen.

This entry was posted in ClimateBall, Global warming, Policy, Roger Pielke Jr and tagged , , , , , . Bookmark the permalink.

55 Responses to Plausible emission scenarios

  1. Nick Palmer says:

    It seems to me that there’s a little ‘unrealistic extrapolation’ – a very useful phrase when commenting on prognostications in many areas – inasmuch as the US’s ‘dash to fracked gas’ and away from coal, and a similar move away from coal in some other major developed nations, has created a bit of a fool’s gold situation whereby moves which were made for purely financial reasons are extrapolated forwards over the next few decades in which the motivation will necessarily have to be ‘save the climate’ rather than ‘save big bucks’ – two different forces.

  2. Actually, I’ve just recalled that one criticism of RCP6 is that it as a rather odd emission profile in that it’s initially lower than RCP4.5. I wonder what impact this has on the analysis in this paper.

  3. Nick,
    Yes, I did also think that was a bit too depending on extrapolating from current trends. They could argue that the scenarios themselves are somewhat based on attempts to develop realistic storylines. However, as I mentioned in the comment above, RCP6 actually has lower emissions to mid-century than RCP4.5. So, it seems possible that they’re excluding it because the emissions are growing too slowly, rather than too fast. So, it’s not clear that all the emissions scenarios are actually realistic (i.e., you’d probably expect RCP6 to grow faster than RCP4.5).

  4. Chubbs says:

    Wow, what a weak paper. Twelve year old scenarios aren’t perfect. Who could have guessed? Renewables and natural gas have out performed even the most optimistic forecasts making the CMIP5 fossil fuel scenarios too pessimistic. However. renewables and fuel switching aren’t even mentioned in the paper; and, unfortunately, emissions ignored by the paper: land use change, methane, and N2O are all on RCP85 trajectories.

    I can buy 2.5C as plausible, but for different reasons than those cited in the paper: 1) Non-fossil technologies are out-competing fossil fuels, 2) Climate change is becoming increasingly obvious and will only worsen going forward, 3) Climate policy isn’t that onerous after all.

  5. Chubbs,
    Yes, my general view is that if we really already done enough to limit warming to ~2.5C, then presumably we can do better and get to < 2C, or even < 1.5C. Of course, without negative negative emissions we can't reverse any warming, but either (as you say) climate policy isn't as onerous as people think, or we haven't been quite as successful as some of these analyses suggest. Of course, it is also probable that it gets increasingly difficult. It's probably much easier to go from a ~4C trajectory to a ~3C trajectory, than it is to go from a ~3C trajectory to a ~2C trajectory.

  6. Chubbs says:

    There have been papers which compare the progress of key technologies vs CMIP5 projections. Renewables are growing as anticipated while stack carbon removal and nuclear are floundering. If we are going to make 2.5C or better there is only one horse left to ride

  7. Nick Palmer says:

    Chubbs wrote “while stack carbon removal and nuclear are floundering”

    I think that’s more of an assertion than a fact. We’ll almost certainly need all those as well as renewables. That Mark Z Jacobson paper purporting to show that the job can be 100% done with renewables has been heavily criticised/debunked by such top climate scientists as Ken Caldeira.

  8. paulski0 says:

    I was going to post this on twitter, but it’s quite long so maybe putting here would be better:

    A few comments…

    These short-term trend tests in series with decadal variability aren’t robust, particularly given the very narrow thresholds used. If you test 2000-2015 emissions trends instead of 2005-2020 using the previous RCP marker generation then it finds RCP8.5 is the only “plausible” scenario.

    The 2020-2040 test is comparing an energy system model with other energy system models. How is inherent “plausibility” established here? IEA STEPS uses a single model with no uncertainty analysis, following a single future policy scenario and a single set of socio-economic forecasts. It’s then being compared to different models following different policy scenarios and different socio-economics. Perhaps the authors should consult this paper, which goes into detail about how this is bad practice 😉

    The IEA stated policies scenario being based on real-world policy plans/commitments confers “plausibility” of course, but resultant emissions are also a function of the particular model and underlying socio-economic assumptions. For example the SSP2 baseline runs of the IMAGE and MESSAGE-GLOBIOM models differ in emissions growth rate by more than 0.3%/yr over 2005-2020 and 2020-2040, and that’s just through being different models with the same socio-economics. This also returns to my first point about the robustness of these tests.

    Really what needs to be done is for equivalent stated policy scenarios to be run in the SSP models with varying socio-economics to show a resulting distribution of emissions. But that would kind of reveal the tautology of the paper, since it’s ultimately just saying that if we follow a scenario of incrementally increasing policy then we won’t be following scenarios where that doesn’t happen.

    Finally they extrapolate to find 2100 warming based on identified “plausible” scenarios. This has the same problems as I discussed recently regarding other attempts to extrapolate based on current/stated policies – that in most scenarios the vast majority of policy enacted in order to achieve strong climate targets occurs after 2040, so what happens up to 2040 has limited relevance. For example, Pielke et al. suggest 3.4 being the most plausible scenario level. The median indicative carbon price across 3.4 scenarios in the SSP database is $56 at 2040, and then $816 at 2100. Are the authors really arguing that a 15x increase in policy strength between 2040 and 2100 is simply a fait accompli assuming stated policies are implemented by 2040?

  9. Paul,
    Thanks. That’s a good point about it not being robust to choice of time interval.

  10. notabilia says:

    On the other side of the statistics side, Peter Carter documents how the real stats are screaming RCP 8.5 in this video:

  11. FWIW, I think it’s pretty clear that we are moving away from RCP8.5 and that if we do continue to strengthen climate policy (and given that alternatives are getting cheaper) will continue to do so. I’m less confident that we can completely rule out some of the high emission pathways than some seem to be.

  12. Willard says:

    > Perhaps the authors should consult this paper, which goes into detail about how this is bad practice


  13. > Perhaps the authors should consult this paper, which goes into detail about how this is bad practice


    Yes, there is something ironic about publishing one paper suggesting scenario misuse is a research integrity issue and then publishing another paper that doesn’t appear to address all the caveats, assumptions, and uncertainties when discussing the plausibility of scenarios. As an aside, Roger seems a bit upset on Twitter because I said “research misconduct”, rather than “research integrity”, but I’m not sure I see all that much of a difference (Research misconduct is typically handled by university Research Integrity offices).

    There’s also a possible conflict of interest here. Roger wants to be an observer who gets to comment on how other researchers are using scenarios, implying that some are misusing them, while also being a research who directly considers the plausibility of scenarios. I can’t think of another example where a researcher both presents an analysis while also presenting a different piece of work essentially suggesting that other researchers who may not agree with their analysis are lacking research integrity.

  14. Willard says:


    I think the main point that needs to be hammered home is the following Climateball rule of thumb:

    (RT1) Mr. T is nobody’s friend, and luckwarmers even less.

    The more uncertainty we have, the less accurate central estimates become as measures of plausibility. Or to borrow from the Auditor, don’t try to argue in one piece that we can’t shoot straight and in the other brag that you hit the bull’s eye.

  15. don’t try to argue in one piece that we can’t shoot straight and in the other brag that you hit the bull’s eye.

    Indeed. There’s also the issue that scenarios are meant to be plausible and consistent descriptions of the future while refraining from ascribing some kind of probability. Of course, if you want to do some kind of cost benefit analysis you need some kind of reference, and I think it’s probably reasonable to think about what might, or might not, be likely given where we are now. But I do think it can be problematic to suggest that some scenarios are more plausible than others without making very clear that they’re still conditional on a lot of assumptions about the impact of current policies and the implementation of future policies.

  16. Also, this is not an independent system. If we suddenly believe that we’ve done enough to limit warming to 2.5°C and relax efforts to limit future emissions, the higher emission pathways become more plausible. On the other hand, if we believe that we should re-double our efforts to limit future emisisons, then the lower emission pathways become more plausible.

    This is a really nasty aspect of all this.

    Unless damages are (perceived as) large and/or early “enough”, we will get more tepid mitigation. Yet, on the other hand, even the tepid mitigation we have had to date seem to be reducing the *probability* of the kind of (this century) damages that would stimulate the actions needed to really drive down emissions steeply and early.

    Since even radical emissions reductions have very little impact to the delta of experienced temperature increase/worsening damages for the next, say, 25 years, the “carrot/siren call” of “we seem to be on course to avoid the severe case damages” can easily induce a slow walk on mitigation.

    The late Marty Weitzman used to, and Ken Caldeira currently talk about this. Not specifically in terms of scenario plausibility, but in terms of human urgency and agency in a system where the global temperature is changing ~0.02°C /yr. Barely perceptible as we slow walk through it in real-time.

  17. Chubbs says:


    My take on the lack of progress in nuclear and CCS vs expectations in climate scenarios is taken from the paper below. The paper is several years old, but the story hasn’t changed. Per IEA and BP 2020 energy outlooks, if we want short-term emission reductions, heavy lifting from renewables can be expected.

    Click to access BudgetCommentary2016_20161210.pdf

  18. b fagan says:

    Since you mention “find some way to extract hydrocarbons from clathrates”:

    “Energy Resources from the Bottom of the Sea
    Research is underway into the commercialization of natural methane hydrate believed to exist in the seabed layer in the waters surrounding the Japanese archipelago.”

    And from Japan’s Ministry of Economy, Trade and Industry from early 2019:

    “METI Revises the Plan for the Development of Marine Energy and Mineral Resources”
    The revised development plan specifically defines: approaches to research and development from the first step to goal achievement by type of mineral for marine energy and mineral resources and technical development and other necessary related elements; and it also defines the future direction of plans for developing such resources for the coming five years or so. Under the plan, METI will continue to make efforts toward achieving these goals.

    Methane hydrates:

    Regarding the pore-filling sand-layer type, METI will conduct: development of technologies for long-term production, onshore production tests, research and test drilling in the sea areas surrounding Japan, surveys on marine environments and other efforts.
    Regarding the shallow type, METI will assess the results of the review of the literature on current potential recovery technologies, undertake research and development of necessary recovery and production technologies and undertake surveys on states of the seabed, surveys on marine environments and other efforts.”

  19. Pingback: Eight years | …and Then There's Physics

  20. Willard says:

    I see that Junior remains a charming Climateball chap:

  21. Person 1: If we’re not careful something terrible could happen.
    Person 2: I don’t think that such an outcome is plausible.
    Person 1: Yes, it may not be very likely, but I still think it’s possible if we’re not careful.
    Person 2: Do you want to bet on it? I don’t think it’s plausible, so I’ll bet that it won’t happen.
    Person 1: No, I don’t really want to bet on an outcome that I don’t think is all that likely and that I’d rather we were careful to avoid.
    Person 2: Well, if you won’t bet on it, then I think that implies that you don’t really think it’s plausible.
    Person 1: What?

    Of course, the other narrative could be:

    Person 1: If we’re not careful something terrible could happen.
    Person 2: I don’t think that such an outcome is plausible.
    Person 1: Yes, it may not be very likely, but I still think it’s possible if we’re not careful.
    Person 2: Do you want to bet on it? I don’t think it’s plausible, so I’ll bet that it won’t happen.
    Person 1: Sure, I’ll take that bet.
    Person 2: Really, you’re willing to bet that some terrible thing will actually happen? Benefitting from other people’s tragedy?
    Person 1: What?

  22. Willard says:

    After all these years, Junior does not realize that betting on weather is fun because it’s just weather.

    Speaking of which, Pierre paid his due:

    Also note the difference between “I believe it will warm” and “I believe that P is more plausible.” That it will warm is easy to verify, whereas the fact that you rolled a double six with two dice does not imply it was the most plausible outcome.

    Worse still is that nobody should bet on the *possibility* of rolling a double six if that implies something Very Bad for our children and their children.

  23. I guess some will remember the classic discussion between James Annan, Roger, and a hypothetical undergraduate called Megan. Unfortunately, some of the posts no longer exist, but you can find them through the waybackmachine if you would really like to.

  24. paulski0 says:

    Samantha and Miranda are in a plane waiting to do their first skydives. Miranda says she’s worried and doesn’t want to jump. Samantha tells her it’s really very safe, accidents are unheard of and she’ll 100% guarantee that Miranda will be fine if she jumps now.

    With that said, Miranda steels herself, clenches her fists, walks to the door and throws herself out, plummeting to her death.

    Many years later Samantha dies and the two meet up in heaven. Miranda immediately says “What the hell happened? You promised me I’d be fine.”

    Samantha grimaces and replies “Well, I did kind of assume you’d put on a parachute first.”

    I’m just not sure how talking about plausibility in this context is useful when the outcomes are entirely up to our decision making.

  25. paulski0 says:

    Regarding the 6.0 thing, I don’t think it has anything to do with the CMIP5 RCP6. They analysed the 6.0 scenarios from the SSP database.

    Roger’s labeling of 6.0 scenarios as implausible based on the IEA STEPS scenario really misunderstands the nature of these scenarios, though to be fair is not dissimilar to how other mainstream researchers have used the scenario databases.

    Mitigation scenarios involve a pre-defined climate target (e.g. stabilisation at 6.0W/m2 forcing) and modeling groups are charged with finding an optimal policy pathway to hit this target given a particular techno-socio-economic configuration (e.g. SSP4). Optimal here simply meaning lowest possible cost to 2100 GDP relative to the no-policy baseline. And all of this under the idealised assumption of zero cost from climate change.

    Because the 6.0 target is relatively easy in most cases compound growth means the optimal path is typically found to involve delaying most mitigation until later in the century. Most 6.0 scenarios in the SSP database are subject to only very low levels of climate policy by 2040, and consequently are not too dissimilar from no-policy baselines at that point.

    It’s therefore not at all surprising that a scenario of relatively substantial policy increase by 2040, such as IEA STEPS, would have lower emissions at that point than scenarios with very little policy. But this does not imply that hitting 6.0 level forcing is implausible since the timing of the emissions profile is really a function of the optimisation procedure, not a fundamental requirement for hitting 6.0 forcing. If policy is not substantially strengthened beyond 2040 emissions could stay level or even drift upwards a bit and reach 6.0 forcing level.

    It is perhaps a reasonable assumption that, having strengthened policy faster than 6.0 scenarios up to 2040, we would continue strengthening faster after 2040. In fact, I tend to think (or hope) there will be at least a few more rounds of strengthening prior to 2040. But it is just an assumption, about the future.

  26. Paul,
    Yes, I agree that the there isn’t really a strong argument for assuming that the emissions till 2040 somehow define what happens beyond 2040, especially if a lot the policy implementation is expected to happen post 2040 (and also especially since the 2020-2040 emissions are still modelled, rather than observed/measured).

    Regarding the 6.0 thing, I don’t think it has anything to do with the CMIP5 RCP6. They analysed the 6.0 scenarios from the SSP database.

    According to the paper they used both the AR5 and SSP scenarios. So, I agree that the 6.0 issue probably doesn’t apply to the SSP scenarios, but might to the AR5 ones.

  27. paulski0 says:


    According to the paper they used both the AR5 and SSP scenarios.

    They’re referring to the AR5 scenario database, which isn’t the same thing as the RCPs. Some of the marker RCP scenarios are in there, but I don’t think RCP6 is.

  28. Paul,
    Thanks, I hadn’t realised there was a difference.

  29. Pingback: What I “read” this morning on my morning walk: 2021-04-14 – Ken MacClune

  30. raypierre says:

    Even if Pielke et al are right, the discussion ignores the fact that the world doesn’t end in 2100. On the optimistic emission trajectory, we wind up in 2100 with higher emissions rate than today, whereas net zero is what we need. The world continues to warm, to 3c, 4c, 6 c and onward in the years afterward.

  31. Ray,
    Indeed, and that has been pointed out to some of those who are claiming that we’ve now ruled out some of the high emission pathways. It doesn’t seem to do much to temper their enthusiasm.

  32. paulski0 says:


    Perhaps I’m misunderstanding you, but that doesn’t seem right. Pielke et al. are claiming that 3.4 level scenarios are the most plausible, resulting in about 2.2C warming at 2100. Across all 3.4 scenarios in the SSP database the average warming rate from 2090 to 2100 is basically zero, and net CO2 emissions at 2100 average basically zero. We wouldn’t expect any more warming after 2100 if we were to follow a 3.4 scenario.

  33. paulski0 says:

    By the way, would I be right in thinking this is technically supposed to be the same paper as the “research integrity” one?

    As far as I know that was never officially published and I’m sure I saw Pielke say that this is a revision to something he’d posted up before.

  34. Paul,
    You may have a point about Pielke’s paper, but some of the other analyses that have suggested that the high emission pathways are ruled out have not ended up with ~zero emissions by 2100.

    By the way, would I be right in thinking this is technically supposed to be the same paper as the “research integrity” one?

    No, I think this is a different paper that has, or soon will be, submitted.

  35. Chubbs says:

    The chart below, pinned on ZekeH’s twitter, is the main reason 2.5C looks more plausible today vs 10 years ago. Didn’t have to turn out that way. But, those technologies have thrived more than anticipated under mild climate/industrial policy. I wouldn’t bet against them in the future.

  36. paulski0 says:


    Scenarios generally expect big price declines for low-carbon tech even under no policy. I have seen suggestions that solar, wind energy prices are well below those anticipated in IAM runs, but it’s not clear that this is actually translating to greater overall low-carbon electricity generation than expected by those IAMS.

    To be specific, by low-carbon electricity generation I mean wind+solar+geothermal+hydro+nuclear+biofuels. I make it that the IEA analysis for 2019 found 36EJ low-carbon generation, BP Stats review found closer to 35EJ.

    Looking at SSP2 scenarios the median 2020 low-carbon electricity generation under no-policy baselines is 36EJ, under SSP2-6.0 runs 37EJ, under SSP2-4.5 runs 37EJ. This says a couple of things: Overall growth in low-carbon electricity projected by IAMs under SSP2 has been closely in-line with observed growth over the past decade, even with no policy. The difference in growth to this point between mitigation levels is very small, mostly because policy difference to this point is very small.

    If we look further ahead the IEA stated policies scenario projects 80EJ low-carbon electricity generation at 2040 by my calculations. At that point the SSP2-baseline median is 58EJ, SSP2-6.0 70EJ and SSP2-4.5 78EJ. Now we can see the difference being made by policy in the scenarios. IEA stated policies projects basically the same low-carbon electricity as SSP2-4.5 by 2040, which makes sense because the levels of policy between the two appear to be generally very similar to that point.

    Despite many suggestions that clean energy tech is developing faster than modeled on a policy-for-policy basis, observations to-date don’t back that up, and current forecasts to 2040 don’t back that up. Why this apparent disconnect? There may be a few reasons but I think the largest is that this idea of rapid clean energy growth is pretty much always represented by just solar and wind, and increasingly just solar by itself, because those are the fastest growing technologies. Those that aren’t doing so well, such as nuclear, are left out of such presentations despite being massive sources of clean energy. But the thing is, it doesn’t matter for the climate which technologies win out. What matters is the overall amount of clean energy generation, so it doesn’t work to just cherry pick the fastest growing technologies at any one time and imply that they’re representative of the whole.

    I’m not suggesting you’re doing this, Chubbs. It’s just a common thing which seems to happen and bugs me.

  37. Willard says:

    > it doesn’t matter for the climate which technologies win out

    Considering the amount of energy we’ll need, it is to be hoped that every tech finds a way to win …something.

    Do the SSPs guesstimate the plausibility of the capacity they presume regarding renewable tech?

  38. paulski0 says:


    Do the SSPs guesstimate the plausibility of the capacity they presume regarding renewable tech?

    In a sense, perhaps. SSP2 is explicitly the median scenario so, assuming a normal distribution, could be reasonably interpreted as the most likely socio-economic pathway (the obvious caveat being “most likely as far as we know, which isn’t a great deal”).

    SSP1 and 3 are explicitly designed as upper and lower cases for socio-economic development so could reasonably be interpreted as having much lower plausibility in that normal distribution. Likewise SSP5 is basically depicting an upper socio-economic development world with very low focus on sustainability, so can be regarded as having much lower plausibility. SSP4 is an interesting one – sort of a jumbled up SSP2 remix where the rich get richer and the poor get poorer – and probably the closest scenario to what’s happened in the real world over the past decade.

  39. Ben McMillan says:

    Well, if you blob all the low carbon stuff together, then it looks like nothing dramatic is happening recently. Wind+solar is now comparable to nukes, and considerably less than hydro, so doesn’t stand out.

    The reason that people are excited about wind+solar is that they are doubling every 5 years or so at the current pace, and look like already lowest-cost most places. Getting excited about an exponential before it overwhelms everything else looks sensible from my perspective.

    The implied assumption is that solar+wind will get so big they will become the main clean energy story; this may be wrong, but I don’t think past trends can tell you one way or the other.

    Like electric cars. Probably not impacting oil use much at the moment. But I think the oil companies are looking a bit further ahead.

  40. paulski0 says:


    It really shouldn’t be surprising that some technologies develop faster than expected, some technologies slower, given expectations set at some shared point in time. It perhaps wouldn’t be that surprising for overall clean tech to develop faster either, but the evidence we have is that this isn’t happening so far – the fast and slow are balancing each other, and appear set to do so for the foreseeable future.

    The issue is that some people are presenting a story that scenarios are substantially under-predicting clean energy growth, and therefore presumably overestimating future climate change, by only showing the fast growing clean energy tech and ignoring the stalled and slow growing tech. In reality there’s no clear evidence so far that scenarios are underestimating overall clean energy growth.

    The implied assumption is that solar+wind will get so big they will become the main clean energy story;

    Well, yes. But that would also imply that other clean energy technologies fall by the wayside. Some scenarios expect/expected nuclear to go exponential as a dominant clean energy source, which looks unlikely to happen now. Great if we get 50EJ more solar than expected by 2100, but if we get 50EJ less nuclear we’re in the same place climate-wise. And these aren’t independent factors. It’s not a totally zero sum game, but the stalling of nuclear is partly being caused by the rise of solar.

  41. Ben McMillan says:

    Well, I’m certainly not arguing the scenarios are underestimating current clean energy growth. But to me, this looks like something happening, that may be important going forward:

    I guess the point is just that if wind+solar really did continue doubling every 5 year, stalling of other clean tech would be pretty irrelevant. That trend seems unlikely to continue very long, but views of the foreseeable future differ (which surely means it isn’t foreseeable?). And Chubbs’ point on cost is relevant: clean tech cheaper than fossils really isn’t something expected in most scenarios, and this is grounds for some optimism.

  42. paulski0 says:

    For completeness/further illustration I’ll post individual projections for solar, wind, nuclear electricity generation comparing SSP2 scenarios with IEA Stated Policies at 2040:

    IEA: 19.7 EJ
    SSP2-Baseline median: 5.6 EJ (1.5 – 9.6 EJ)
    SSP2-6.0 median: 7.3 EJ (1.7 – 11.3 EJ)
    SSP2-4.5 median: 9.3 EJ (2.3 – 18.8 EJ)

    Solar is set to comfortably outstrip general expectations. It should be pointed out that one of the six models produces about the same solar growth under SSP2-4.5, so it’s not entirely beyond SSP modeling but the next highest is 10.9 EJ so that’s a big outlier.

    IEA: 19.6 EJ
    SSP2-Baseline median: 13.5 EJ (4.0 – 18.9 EJ)
    SSP2-6.0 median: 17 EJ (10.6 – 36.3 EJ)
    SSP2-4.5 median: 19 EJ (12.2 – 46.4 EJ)

    Wind is on-track to closely follow SSP2 median expectations on a like-for-like policy basis. It’s really only solar which is growing faster than previously expected. Wind now seems to be following a steady linear growth path rather than exponential.

    IEA: 12.4 EJ
    SSP2-Baseline median: 18.8 EJ (7.1 – 25.2 EJ)
    SSP2-6.0 median: 24 EJ (7.6 – 27.3 EJ)
    SSP2-4.5 median: 26.3 EJ (11.3 – 32 EJ)

    Here we can see that SSP scenarios expected a big increase in nuclear generation, which doesn’t seem to be materialising. In fact, the median shortfall relative to SSP2-4.5 is larger than the solar surplus. Hence why overall clean tech is not looking like outperforming SSP modeling expectations.

  43. Chubbs says:

    Thanks Paul. I am not familiar with IAMs or scenario details. Agree that slow nuclear growth could offset the benefit from solar/wind. Would echo Ben’s point though. Solar/Wind/Batteries/electrification combined have large potential in the future to displace fossil fuels. Fortunately solar is very amenable to acceleration with the right policies, as discussed in the twitter thread below. Also note difficulty in modeling solar.

  44. Ben McMillan says:

    A reminder of the IEA’s accuracy predicting solar:

    IEA stated policy 2020 is still claiming that the rate of installation of new solar/wind will basically be flat from now until 2040. It is just that the current rate of solar addition is so large that this now looks pretty big.

    I.e. there are strong grounds for thinking that the IEA might be too pessimistic on renewables (like they have been for the last two decades).

  45. Mr. Hoekstra, if you push the graph back to 1978, solar’s growth is even more impressive. Long may it continue.

  46. Tom,
    I don’t think Auke is a regular reader of this blog.

  47. Willard says:

    > I don’t think Auke is a regular reader of this blog.

    Damn. That makes me realize that I don’t have my Cowbell Theory of Truth thread online anymore. I know you’re not reading, Twitter Support, in fact you don’t even read the emails we send you, but thanks anyway.

    I still got it in my archives somewhere.

  48. Ben McMillan says:

    The big difference that cheap renewables make in terms of plausible scenarios is probably not at 2040, but later in the century. Renewables are just not big enough to make a major dent before 2040 unless you are somewhat optimistic about growth.

    Highest-end scenarios (I think people are usually moaning about SSP5, not SSP2) tend to use up all the easily available oil+gas by mid-century, and are then forced to switch to coal, which is rather inconvenient for anything except electricity generation (i.e. you need to electrify everything or do coal-to-gas/liquids) and not that cheap even if you have a large domestic supply. Coal use increasing by a factor of 5 (mostly after 2040) looks increasingly implausible with new-build renewables already cheaper in most places than new coal.

    What looks very different between current trends and higher-end scenarios is gas usage for electricity; that is meant to triple or more over the next couple of decades. The highest-end scenarios also have coal electricity generation doubling in the next 20 years. But at the moment there is not that much net coal or gas capacity being built out.

    Also: I think you could make a reasonable case that wind installations haven’t been that flat over the last decade.

    Of course, this doesn’t get you to 2.5C scenarios. But it really does make the kinds of scenarios likely to lead to 5C less probable…

  49. Ben McMillan says:

    Oh, and if you take the IEA STEPS seriously, emissions to 2040 will be flat, and thus probably close to SSP2-4.5, which is pretty much a 2.5C scenario.

    And that is the most pessimistic of the IEA scenarios…

    Not that it is mostly to do to with renewables, though.

  50. Chubbs says:

    This 2019 Science paper outlined some high solar penetration scenarios, which start moving the needle after 2030.

  51. paulski0 says:


    The big difference that cheap renewables make in terms of plausible scenarios is probably not at 2040, but later in the century.

    Well, that leads to my question. If solar/wind are really so much cheaper in a genuine like-for-like way, and set to get even cheaper, why wouldn’t fossil fuels be eradicated from the electricity supply by 2040, especially given the assumed policy implementation to make burning fossil fuels even more expensive?

    My suspicion is that the prices typically quoted represent a selection of situations ideally suited to solar/wind installations rather than being representative of costs given an average power supply requirement, and therefore may not be providing like-for-like numbers compared with coal power, or to compare with models. My understanding is that the figures come from costs in completed projects, which means the solar price wouldn’t include projects where solar wasn’t the cheapest, and therefore provides a biased perspective.

    I could be completely wrong about this, but I suspect costs associated with solar/wind power installations would be far more variable than coal power given a range of hypothetical situations, which means if statistics only reflect costs of the winning power source in any situation it’s guaranteed that reported solar/wind prices will be biased low.

    It also seems likely to me that what we’ve done so far with solar and wind is pick the low-hanging fruit. While we can expect technologies to improve and theoretical costs to come down even further, I’d think that will be set against a trend of it becoming increasingly difficult and costly to find viable places to put this stuff.

    Oh, and if you take the IEA STEPS seriously, emissions to 2040 will be flat, and thus probably close to SSP2-4.5, which is pretty much a 2.5C scenario.

    As discussed above, emissions trend comparisons with SSP scenarios to 2040 really should not be interpreted as necessarily indicating much about 2100. The emissions profiles over time in these scenarios are to a large extent effectively pre-defined by the experimental design so should not be used as if they provide an emergent probabilistic guide. Yes, a relatively flat emissions trend to 2040 is part of SSP2-4.5 scenarios, but equally limiting to 4.5 then requires a substantial decline in emissions from 2040 to 2100, by about 50-75%, which is achieved thanks to a sharp acceleration of policy – effective global carbon price increasing by 5-18x from 2040 level. Stated policies as-is would be expected to continue along a roughly flat emissions trend, resulting in about 3C warming by 2100, probably increasing to 4C by late 22nd Century.

  52. Ben McMillan says:

    I guess my judgment is that in most places wind+solar resources are plentiful, so running out of good places to put them is not too big a deal. Solar potential in particular is gigantic. Some bits of Europe have rubbish wind and solar potential but that is generally an exception. Given that even fixed-based turbines can go in water deeper than 50m, there is e.g. massive offshore wind potential in the UK (i.e. MacKay was too pessimistic about this).

    But costs are indeed variable from place to place; solar and wind are starting to dominate new construction, but there are still places building fossil plant (and it isn’t all economics). That sounds to me exactly like what you would expect if on average, wind+solar were now somewhat cheaper than new fossil.

    Even if the last fossil plant is built in 2030, it still runs to 2060 or until new wind+solar is cheaper than running an existing fossil plant. And that’s just electricity.

    On STEPS: sorry, I remembered after I posted that ATTP (in the OP) and others had basically discussed this above.

    In STEPS, wind+solar together are the biggest source of electricity in 2040, and there is already quite a bit of electrification. Do we really figure out how to get half-way toward zero-carbon tech, but it never really catches on? I think the idea that we will need extremely strong ‘policy’ to reduce carbon emissions in the last half of the 21st century is a bit silly. e.g. electric cars, powered by renewables, are already competitive with ICE vehicles. People driving around gas cars or burning things to heat their buildings in 2100 is hard to believe (or requires some kind of Mad Max scenario) if we already get half-way by 2040.

    I think the transition has ‘momentum’, so a flat line in emissions is a bit odd.

    More likely to me at least is that the transition could take a long time like in STEPS, because fossil infrastructure is locked in, and inefficient housing will take tens of years to replace. But once underway, we will eventually entirely dump fossil fuels.

    I think IEA’s emphasis on ‘policy’ is the reason its forecasts on renewables have been so spectacularly wrong. Everything looks like a straight line in their scenarios, because their assumptions on technological change are so conservative. The transition is already happening regardless of policy, the remaining question is how fast.

    (and if we are talking about whether you end up at 3C instead of 2.5C at 2100 that is still a long way from the worst-case SSP5 scenarios)

  53. paulski0 says:


    Greater recent solar+wind new construction is not necessarily indicative of a current market lead, there are a number of other factors. One is the business cycle. During the real boom years (say, up to about 2015) China were building massively over capacity, in power plants and many other things. As things have slowed, they’ve been filling that capacity rather than building much more, hence why Chinese cement production has been in decline over the past several years. Despite a big slow down in new builds, Chinese coal electricity generation growth has massively outpaced any other source over the past few years. Globally, absolute coal+natural gas electricity generation growth has continued to outpace solar+wind in recent years, by about 2x.

    There’s also influence from policy on recent building. Not just current policy, but the threat of future policy. A number of fossil fuel projects have been cancelled citing expectations for future strengthening of climate policy, which is in an interesting phenomenon. I wonder if/how the models deal with that.

    Do we really figure out how to get half-way toward zero-carbon tech, but it never really catches on?

    Yes, zero-carbon is about 50% of electricity generation in STEPS at 2040. But then, the figure now is 35%. We’re not talking about going 0-50. It’s a reasonably quick, but still steady transition. Flat emissions does not imply a halt to that transition. SSP2-6.0 is probably the closest scenario to an extended STEPS over the whole period to 2100. The median zero-carbon fraction of electricity generation at 2100 across the six models under that scenario is 75%, so there is a continued move towards these technologies. However, total electricity demand also goes up so the absolute amount of fossil fuel electricity generation remains roughly constant.

  54. Ben McMillan says:

    I’m not sure how you are getting gas+coal generation (together) growing faster than wind+solar (together), unless you are looking over timespans longer than 5 years (solar is basically negligible 5 years ago and even wind has sped up over the last decade)

    Business cycles in individual countries seems to be getting into the weeds a bit.

    I think people ‘seeing the writing on the wall’ is probably a big part of what is happening, despite relatively weak policy in terms of government regulation. I guess business culture and individual preference would get lumped with ‘policy’ in an IAM analysis. Some of that writing on the wall is the prospect of further tech change though: assets can be stranded both because of regulation and because of being rendered uncompetitive.

    I was overstating how much STEPS is a progressive scenario, but if all the energy growth is in low-carbon technology for 20 years, I would be inclined to think that trend is likely to accelerate going forward. Of course, worse outcomes are possible, which aren’t implausible, just seems unlikely.

    Hopefully though policy keeps tightening anyway and makes this moot.

  55. David B Benson says:

    A survey of many of the passive radiative cooling materials:
    which doesn’t include the newer use of barium sulfate microspheres.

Leave a Reply

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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.