If you’ve been following the blog for a while, you will be aware that I’ve commented on a number of occasions about the whole RCP8.5 issue. You may also be aware that one of the chief protagonists in that whole discussion is Roger Pielke Jr, whose work I’ve also discussed from time to time. If so, you may be interested in his latest attempt to police the climate science community.
It’s a working paper on the Systemic Misuse of Scenarios in Climate Research and Assessment. The other author is Justin Ritchie who, in the past, wrote quite a sensible article with Zeke Hausfather.
I don’t really want to say too much about the new article, as I’m wary of incuring the wrath of Roger. One could, however, play a reasonable game of climate bingo with it. It includes Climategate, cites Grundmann (2013) unironically, of course discusses RCP8.5, implies a lack of research integrity amongst the climate research community, discusses problems with the IPCC and the US National Climate Assessment, and explains how climate research can get back on track (and avoid a growing credibility crisis).
I won’t say much more, as my main goal was to simply highlight the paper for those who might be interested. I’ll end, though, with a response from Nico Bauer, who is an Integrated Assessment Modeller at the Potsdam Institute for Climate Impact Research.
Links:
Systemic Misuse of Scenarios in Climate Research and Assessment – new working paper by Pielke & Ritchie.
https://www.carbonbrief.org/analysis-how-carbon-cycle-feedbacks-could-make-global-warming-worse – new Carbon Brief article by Zeke Hausfather and Richard Betts. I didn’t mention this in the post, but it does seem relevant to the whole RCP8.5/scenarios debate.
...and Then There's Physics 
On a more serious note, I think there are entirely valid criticisms of how scenarios are used in climate research. However, I do think that it’s extremely poor to attempt to publish a paper claiming that a large group of other researchers are “misusing” scenarios and implying that they lack research integrity. For starters, if your goal is to actually get others to engage with your criticism, it’s unlikely to be well received (there are aspects of science that are social, and at least one of the authors should realise this). It’s also extremely pompous (be careful of that high horse).
More fundamentally, though, a lot of this is very value-laden. There are no trivial answers to some of these issues. There are – in my opinion – valid arguments in favour of focussing on worst case scenarios, and valid arguments in favour of focussing on more likely scenarios. One of the issues is the labelling of the various scenarios (no-policy baseline, business-as-usual). These can be problematic, but we do need to use words. In some sense, none of the possible terms are ideal, because there’s no such thing as a no-policy baseline, and business-as-usual is a moving target. Should we kick up a big fuss about this, or acknowledge that no terminology is going to be ideal and aim to be clear about what we mean?
I don’t think that there are any easy ways to establish who is right, and who is wrong, in this context. The authors of this recent article obviously disagree, though (they think they’re right).
Perhaps I can sum up Roger’s approach succinctly?
If only those poopyheads would stop calling us poopyheads we could make some real progress here.
Third and fourth links identical?
Bob,
Thanks. Fourth should now go to the working paper.
ATTP: You might want to append a reference to this article in your OP.
Analysis: How ‘carbon-cycle feedbacks’ could make global warming worse by Zeke Hausfather & Richard Betts, Carbon Brief, Apr 14, 2020
https://www.carbonbrief.org/analysis-how-carbon-cycle-feedbacks-could-make-global-warming-worse
JH,
Yes, that is a good article. I’ve added a link at the end of the post.
Neither author are my so-called go to experts on the current AR6 scenarios. Expect to see this published in Energy&Fuels …
https://pubs.acs.org/journal/enfuem
… or some such energy related journal.
Someone on teh Twit suggested that their title for said paper amounted to an accusation of fraud by the scenario developers, an in their choice of the first two words “Systemic Misuse … ”
If you are a climate scientist or otherwise please don’t help them on this one, let the deniers pile on instead, their contributions will ensure that this paper only gets published in E&E or Nature Scientific Reports
I might bother with it somewhat when actually published in a so-called peee-reviewed journal.
“There are also other types of feedbacks that can affect emissions that are not currently included in models. For example, a 2019 paper by researchers at the University of California, Irvine looked at economic feedbacks in high-emission scenarios – in other words, how economic damages from a high warming world might slow growth and emissions.”
yup
Steven,
It does seem that socio-economic feedbacks could limit how much our economies grow and – hence – how much we emit. Some use this to argue against using RCP8.5. My own view is that it implies that things will get pretty bad before we get to RCP8.5. I don’t think it implies things are somehow better than we expect.
Roger Pielke Jr.’s (RPJ) views on projections and predictions don’t seem coherent. For example, back in 2008 when writing in the Climate Audit blog comments, he rejected the idea that Hansen et al.’s 1988 (H88) model-based forecast was right, along with other IPCC forecasts [1]. RPJ doubled down on that last December in the Real Climate blog comments, affirming that H88’s forecast was wrong, and that Gavin Schmidt was wrong to claim otherwise [2]. Yet when writing in Forbes, RPJ cites Schmidt to affirm that climate model forecasts were right [3; 4] (which would include H88’s forecast being right). So RPJ contradicts himself between what much of the public sees him write in Forbes, and what much less of the public (but a number of climate scientists) sees him post in blog comments.
The problem seems to be that RPJ selectively chooses when to acknowledge a particular point: that climate model physics (including shorter-term climate sensitivity) is not about predicting human emissions. So to assess model physics, one should adjust for differences between observed forcings vs. model-projected forcings that result from differences between observed emissions vs. model-projected emissions. Schmidt et al. did that in their recent paper [5], which RPJ cited in his Forbes piece [4]. So RPJ tacitly accepts this point when writing to a larger audience in Forbes, but seemingly disregards it when provoking climate scientists on climate blogs. That’s what enables him to say model-based forecasts are right to the former audience, while saying the opposite to the latter audience. I wonder which choice he’ll make in his future work on projections?
1 : http://archive.is/gRokW#selection-5113.0-5117.236
2 : http://archive.is/XdXrB#selection-891.1-947.237
3 : http://archive.is/lHM9D#selection-2019.0-2027.189
4 : http://archive.is/sMnNR#selection-1815.15-1819.1
5 : “Evaluating the performance of past climate model projections”
Atomsk,
Roger seems to often contradict himself. I do remember the Hansen 1988 issue and I did find that rather bizarre. Does Roger really not get that climate models are not trying to predict the actual emissions and that if you want to check how good a model is, you need to account for the difference between the emission scenario used as input to the models and the emissions that actually happened? It’s possible that he doesn’t get this and that this is driving his criticism of scenario use. It’s also possible that he does but he thinks that many do not and so will see his criticism as making some kind of sense.
From my discussions with him on Twitter, he clearly didn’t understand that the scenarios are inputs to the models.
Gave it a try; but can’t imagine anyone getting through a small fraction of the 63p. Here is my attempt to simplify: the IPCC/climate scientists can’t predict what people will do in the future. True, of course, but one wonders if climate isn’t like an epidemic – the best and worst modeling outcomes don’t verify.
Make take:
Both Roger Sr and Roger Jr crave attention and will keep the pots boiling in order to get it. Both have oversized egos. In other words, they are miniature versions of the Pretend President. The best way to deal with them may very well be to ignore them.
Chubbs:
Man-made climate change degrades the global biosphere and impacts all living things possibly except for the micro-organisms living deep below the Earth’s surface. Pandemics do not, even come close. The modeling done for each are light years apart.
https://twitter.com/nevaudit/status/1253691076994572288
On the plus side, who has got time to read a 63 page document full of innuendo and vague accusations of fraud that takes 10 pages to get to the point? (actually, that was just where I got bored).
I think I’ll wait for the one sentence drive-by that this paper will receive in AR6 WGII (or would that be WGIII). It certainly isn’t the type of paper that would be mentioned as part of WGI The Scientific Basis as all the scenarios are, by now, already locked in place. They are crying over spilled milk as it were, with much emphasis on the crying part.
JH – Yes, climate and pandemic models are completely different. The similarity I was trying to convey, is the difficult task both have in projecting human behavior and the fact that the availability of modeling results could change future behavior and ruin the projection. In both cases, we appear able to avoid the worst case without threatening the best.
It’s certainly meandering. Underpinning many of the arguments there seems to be a demand for what is really an illusory sense of precision given the associated uncertainties with this kind of work, plus a lack of appreciation for practical realities.
For example, there’s a whole bit pointing out that different 2.6W/m2 scenarios have different calculated net radiative forcings at 2100 (with differences as much as 0.5W/m2!!! For reference, the stated uncertainty of observed present day forcing in AR5 is greater than +/- 1W/m2). The suggestion of the paper is then that the RCP2.6 pathway should not be considered representative of anything other than the specific scenario that was used to create it. This is just ridiculous to me, particularly considering the only other options available (until very recently) have been RCP4.5, 6.0 and 8.5. Clearly, if people are interested in the climate implications of those RCP2.6-level scenarios as simulated by GCMs then the correct choice is RCP2.6.
A one-sentence summary of the paper could be: “We should do everything we can to make perfect the enemy of good.”
Then there’s another part expressing dismay that the mitigation scenarios used to create the RCP2.6, 4.5 and 6.0 pathways all stem from medium baselines around 7W/m2, rather than a range. But this is missing the whole point: you’re not supposed to be using only the scenarios used to build RCPs. If you want to consider the implications of a wider range of baseline scenarios, find or build those scenarios yourself, find the radiative forcings and use climate model simulations from the most representative RCP pathways for those different forcings in order to understand the climatic implications. That’s the whole idea and it really is not that complicated. Why is this so difficult for some people to grasp?
Chubbs: In both models, human behaviors are input, not output.
paulski0:
Why is this so difficult for some people to grasp?
I many cases, it’s because they haven’t done their homework and consequently have an extremely shallow understanding of the subject matter. “A little knowledge is a dangerous thing.”
Paul,
Indeed, and – in fact – a few individual’s own value-laden perspective as to what the perfect would actually be.
“Steven,
It does seem that socio-economic feedbacks could limit how much our economies grow and – hence – how much we emit. Some use this to argue against using RCP8.5. My own view is that it implies that things will get pretty bad before we get to RCP8.5. I don’t think it implies things are somehow better than we expect.”
the point is economic damage is a negative feedback. when and how it will impact
emissions is unknown. I suggest unknowns be explored.
If I actually wanted to find out if RCP8.5 is plausible I would recommend more Trump.
https://www.drillednews.com/post/the-climate-covid-19-policy-tracker
“The Climate Rules Being Rolled Back During the COVID-19 Pandemic
The Drilled News Climate & COVID-19 Policy Tracker is keeping an eye on climate change-related rollbacks by the Trump administration and state governments amid the coronavirus crisis, along with favors to oil and gas, and other energy and climate-related industries.
…
Drilled News has a running tally of all the different ways the industry is trying to capitalize off of the coronavirus crisis, a list that has totaled about 60 different environmental rollback measures as of mid-April.”
Just 60? That’s way too optimistic. Wait till SCOTUS rules CO2 is plant food.
Stephen,
Yes, of course. I wasn’t suggesting otherwise. I wa mostly responding to the suggestions that highlighting RCP8.5 is alarmist. If negative economic feedbacks will prevent us from getting there, then we should start to be alarmed by the impact of RCP6. 😉
” d) maintaining research practices that normalize careless use of scenarios in a vacuum of plausibility,”
Not to mention a vacuum of responsibility.
The usual line oft quoted here is that a scenario is not a prediction.
Well, yes, it is.
aTTP — I find RCP1.9 terrifying enough. For example, all of southern Vietnam floods and I am sure Bangladesh is hopeless. Even with a maximum effort to remove excess carbon dioxide from the atmosphere.
angech,
Just because you say it confidently doesn’t make it true.
Steven Mosher,
the point is economic damage is a negative feedback. when and how it will impact
emissions is unknown. I suggest unknowns be explored.
Interestingly the paper doesn’t rule out a small positive economic feedback, though the uncertainty range appears to be strongly determined by extent of damage to GDP. There appears to be a largely static positive influence from reduced power infrastructure efficiency and a balance of reduced heating/increased ac due to heat, but then overlaid on that a range based on different GDP damage estimates reducing energy consumption. At the low end of about 2% damage the GDP loss does not fully counteract the heat efficiency loss and you get a net positive economic feedback of about 3ppm. At the high end of about 20% damage you get about 200ppm reduction.
My big problem with the paper is that, unless I’ve missed something, GDP damage in their model setup can only reduce total energy consumption and population is only directly negatively affected by climate-induced mortality. They do not appear to take any account of the effects that GDP damage could have on technological development, which then likely negatively affects energy efficiency improvements and renewables (plus presumably might further feedback to negatively effect GDP). Nor do they take into account the effects on fertility rates, demographics and long-term population growth.
As an example, the Riahi et al. 2011 RCP8.5 scenario includes renewables providing about 30% of 1750EJ primary energy consumption at 2100, in large part because of fossil fuel resource exhaustion. The modelling in this paper apparently simply applies a large reduction to the total energy consumption but implicitly assumes that the energy mix remains proportionally the same, which obviously wouldn’t happen. If this Riahi 2011 world is consuming 250EJ less at 2100 it’s plausible that it would all come out of the renewables part of the mix and have no effect at all on CO2 emissions.
“The usual line oft quoted here is that a scenario is not a prediction.
Well, yes, it is.”
angech – No, a projection is a conditional prediction. if A then B. The scenario is the A. There are scenarios that bracket the plausible outcomes. If you make a statement about which scenario is the most (or least) plausible, that would be a prediction.
This is a very fundamental point and implying it the term is being used in a disingenuous way reflects badly on you not the modellers (as usual).
Again, people fail to realize that the RCP’s were developed to specify all forcing inputs to climate models, in order that apples to apples comparisons could be made between them. They were also intended to roughly represent a reasonable range of potential forcings. Each came with a story that described global fossil fuel use and economic activity.
They aren’t really predictions, but give plausible ranges for human response to climate change. Human response is the major uncertainty for near-term climate change.
ATTP thanks for recognition that some people actually think that way.
DM. That is an interesting point, I will give it some consideration.
The other thing I noted recently was that some people are trying to see if the shut down has an effect on CO2 levels and possibly even the climate.
Surely if the consequences are so fraught and long lasting a little dip should make no impression at all.
If it did we could stop any time we liked in the future with no consequences.
Should we even be looking at all if we are convinced that the effects are irreversible short of long term severe measures?
The RCP scenarios would not make sense?
angech,
Yes, the impact of this current situation is unlikely to have a big impact on climate. Also, even if it did, this isn’t really a suitable way to reduce emissions.
angech
“The RCP scenarios would not make sense?”
You can’t make predictions about the consequences of your actions without having scenarios that span the options that you have. This is because the model that predicts the consequences needs input essentially specifying what the relevant features of our actions were.
As I said “If A then B”, if we want to choose a rational course of actions, we have a set of scenarios A_1, A_2, …, A_N which represent each course of action, we estimate the outcomes of those scenarios, B_1, B_2, …, B_N, and then choose the course of action for which the outcome we like the best. The RCPs are the As, not the Bs.
If you think the RCPs don’t represent the spread of likely conditions, fine, make a scenario of your own and run it through a climate model (there are plenty available in the public domain, but for some reason climate skeptics don’t seem to have done anything with them).
“Interestingly the paper doesn’t rule out a small positive economic feedback, though the uncertainty range appears to be strongly determined by extent of damage to GDP.”
Yes.
I will just say this. Long ago (2007) when I first stumbled across climate science 2 things
caught my attention. Models and scenarios. Given my background it is obvious why.
My Impressin is that the SRES approach was the way to do this. The ontologies were correct.
of course with SRES you are still making guesses, but I think it’s important to guess at the right things.
I’d much rather see guesses about population, about energy use, and about energy mix made
and then concentrations and forcing flow from that. And this usually leads you t an understanding of
which if these things is most uncertain. Tells you what to focus your work on
Steven,
But this depends on what is motivates your research. Climate scientists are mostly interested in how the climate will respond to perturbations. By using RCPs as inputs to their models, they’re at least modelling a set of pathways that have some relationship to a possible future reality. If someone else wants to then focus more on population, energy use, energy mix, etc, they can do so and they can also use the cllimate model outputs to see what the climate impact might be. It should even be possible to do simple analyses that introduce some kind of feedback (emissions -> concentrations -> warming -> impacts -> emissions, etc).
In my view, a lot of those who criticise scenario use fail to realise (intentionally in some cases) that it’s probably quite possible to already do what they’re suggesting should be done, even if the way in which the scenarios have been developed and used isn’t necessarily ideal.
angech:
What responsibility, Doc? What are your expectations of scientists? What do you expect from yourself, your family, your neighbors, your country? Just who is responsible for AGW?
IMO, your comment reveals how alien the culture of science is to you. Do you doubt that scientists fool themselves all the time, just like the rest of us? It’s their training and the discipline of their peers that drives their empirical rigor. You can be sure they work at least as hard as you do! Regardless, they don’t answer to you, any more than you do to them. No, Doc: averting the onrushing global climate catastrophe is as much your responsibility as any climate modeler’s. The vacuum is yours to fill!
@-angech
“The usual line oft quoted here is that a scenario is not a prediction.
Well, yes, it is.”
It is a ‘prediction’ in the same way that when throwing two dice a double six is a prediction.
RCP 2.6 and RCP 8.5 are reasonable guestimates of the upper and lower limits of possible scenarios, just as double one and double six are limits for two dice.
Izen, I understand what you are saying, I think but the fact remains that if you put up a scenario you automatically have an outcome which I would call a prediction. That is how the RCP got the numbers after their names.
Regardless.
–
Everyone seems to be so caught up in the fear of Skeptics saying that prediction [for that scenario] was wrong therefore the science is wrong.
Science should not work like that.
Even if Skeptics might.
It should have the guts to say, this is what will happen with this set of factors and this with those.
But it should also have the intestinal fortitude to say we got that wrong if by some mischance the factors were relatively right but the outcome was markedly different.
Two common reasons for errors, missing out some important factor or less likely the science assumptions in some of the factors was wrong. Rinse , Learn, modify repeat.
–
Anyone claiming to be completely right virtually all of the time has some communication problem.
Science is allowed to take wrong turns at times without prejudice.
Angech “ ng, I think but the fact remains that if you put up a scenario you automatically have an outcome which I would call a prediction. ”
It has been repeatedly explained to you why you would be wrong to call it that.
‘ Everyone seems to be so caught up in the fear of Skeptics saying that prediction [for that scenario] was wrong therefore the science is wrong.
Science should not work like that.’
No, nobody fears skeptics not understanding basic concepts or using faulty logic, it is par for the course. We are all quite used to it by now.
‘ Anyone claiming to be completely right virtually all of the time has some communication problem.’
A bit like someone that repeatedly makes the same error, despite the error having been repeatedly and clearly explained to them?
” Everyone seems to be so caught up in the fear of Skeptics saying that prediction [for that scenario] was wrong therefore the science is wrong.”
It would be quite helpful if skeptics were to learn that there is more to science than binary “right” and “wrong”. We are at the stage where what we need to do is to constrain the uncertainties. Of course understanding or productively contributing the science isn’t necessarily their goal.
Incidentally, skeptics can’t claim that the “prediction [for that scenario] was wrong” unless the scenario in question accurately represented what actually happened (which would be an unreasonable expectation).
“If angech would actually read, understand, and learn from all the comments that point out his errors, then he would be much less likely to keep repeating his errors.”
Am I making a prediction that angech is going to stop repeating his errors? Yes, or no.
@dikranmarsupial says:
Re: “[“The usual line oft quoted here is that a scenario is not a prediction.
Well, yes, it is.”]
angech – No, a projection is a conditional prediction. if A then B. The scenario is the A. There are scenarios that bracket the plausible outcomes. If you make a statement about which scenario is the most (or least) plausible, that would be a prediction.”
I made a similar point to Roger Pielke Jr. using a medical analogy: http://archive.is/XdXrB#selection-1287.0-1303.136
I don’t get what some people find so hard to grasp about this. For example, suppose you tell a child: “if you place your hand on the hot stove, then your hand will burn”. The child then doesn’t place their hand on the hot stove, and their hand doesn’t burn. It would be absurd for the child to then claim that you were wrong, since they didn’t place their hand on the stove. After all, your statement was a conditional projection of what would happen under a scenario in which the child placed their hand on the stove, not a prediction of whether the child would actually place their hand on the stove. Yet this is what angech, Pielke Jr., etc. do; they try to claim a conditional projection failed, since they claim a predicted scenario didn’t occur. Amazing. They wouldn’t try this in everyday life, as per the hot stove example above, nor in another branch of science. So why do they keep doing it in climate science?
Or to put this another way: it’s illogical to claim that you falsify an “If A, then B” conditional by falsifying A. You falsify it by showing that B doesn’t follow when A occurs. You can do that indirectly even if A does not actually occur. For example, take the case I previously mentioned where “A” is radiative forcing increase, and “B” is temperature increase for a model-based conditional projection. Even if the precise “A” for the conditional projection did not actually occur, you can compare the conditional’s projected “B/A” ratio (i.e. it’s shorter-term climate sensitivity) to the B/A ratio that actually occurred:
Pingback: Systemic misuse of criticism of climate science | …and Then There's Physics