Extreme weather event attribution

This is a joint post between myself and Eric Winsberg, Professor of Philosophy at the University of South Florida. Eric has just published, together with Naomi Oreskes and Elisabeth Lloyd, a paper called Severe Weather Event Attribution: Why values won’t go away. The paper discusses the issue of how one might assess the anthropogenic influence on an extreme weather event. This post describes what was presented in the paper and tries to justify why there may be value in approaching this issue from more than one perspective.

Extreme weather event attribution

One way that we can gain confidence in our understanding of anthropogenic influences on climate is to carry out detection and attribution studies. The basic idea is to consider an extreme event, or a pattern of extreme events, and to establish the probability of that extreme event occuring under current conditions. This can then be compared to what would be expected had we not undergone greenhouse warming, and other anthropogenic changes, to determine some kind of risk ratio.

One advantage of this approach is that it largely avoids false positives; it will only assign some probability of an anthropogenic influence if there is some clear detection and if some of this can indeed be attributed to anthropogenic influences. However, this also means that there will almost certainly be circumstances where we do not assign a probability of an anthropogenic influence when, in fact, such an influence does indeed exist. From a scientific perspective this might be fine; we would simply be waiting for a sufficiently strong signal to emerge. However, this could potentially lead us to under-estimating the impact of anthropogenically-driven climate change.

A complementary approach is to consider a storyline. For example, given that an event has occured, how might climate change have influenced this event? If the air was warmer, then we may expect enhanced precipitation. If sea surface temperatures are high, then we may expect a tropical cyclone to be more intense. The focus here tends to be on the thermodynamics (i.e., the energy) and to take the dynamics as given (i.e., the event happened).

It turns out, though, that the story-line approach has been rather controversial, with many who favour more formal detection and attribution being highly critical. They argue that it could lead to more false positives and that taking the dynamics as given ignores that dynamical factors could actually work to make some events less likely. Essentially, they argue that the storyline approach may over-estimate anthropogenic influences, potentially mistaking natural variability as being anthropogenic.

The problem, though, is that although the two approaches are complemetary, they’re not actually quite addressing the same issue. The detection and attribution approach is essentially trying to determine how anthropogenic-driven climate change influences the probability of a specific class of event. The storyline approach, on the other hand, is more looking at how anthropogenically-driven climate change might have influenced an event that has actually occured. There is no real reason why we should prefer one approach over the other; they can both play an important role in aiding our understanding of how anthropogenic influences impact extreme weather events.

The criticism of the storyline approach seems to have two main strands. One is that the more formal detection and attribution approach avoids the reputational harm that may occur if climate scientists make claims that later turn out to be wrong. The other, is that the storyline approach involves decisions that are likely to be influenced by value-judgements. Given that the detection and attribution framework relies on probabilities, it may be somewhat closer to value-neutral that the storyline approach, but it’s not completely value-free. There are always going to be judgements associated with things like model assumptions and how to present the results.

Also, the judgement that detection and attribution is preferable to the storyline approach is fundamentally value-laden. It’s a judgement that avoiding false positives is preferable to potentially presenting false negatives. Just as incorrectly associating climate change with an extreme event could lead us to investing in infrastructure that turns out to be unnecessary, under-estimating the link between climate change and extreme events could lead to harm that could have been avoided.

From a scientific perspective, the detection attribution approach may well be preferable. However, the storyline approach seems very valuable from a public perspective. It allows us to consider how climate change may have influenced a specific event and also allows us to discuss how it may impact similar events in future. Without it, you run the risk of articles like this one that uses the lack of a detectable trend to conclude that there’s no solid connection between climate change and the major indicators of extreme weather. The storyline approach would almost certainly not lead to such a conclusion.

Overall, it’s hard to see why we shouldn’t be using both approaches. They’re complementary, address slightly different aspects of the link between climate change and extreme events, and should ultimately tend to be consistent. If an expected influence doesn’t emerge, then we’d have to either re-think the storyline, or double check the detection and attribution analysis. However, the storyline approach can also allow us to stress that the lack of a detectable trend doesn’t necessary imply no link between anthropogenically-driven climate change and extreme weather events; absence of evidence is not the same as evidence of absence.

We would clearly like to quantify the impact of anthropogenically-driven climate change on extreme events, but we would also like to avoid concluding that there is no link when in fact such a link is expected and will ultimately become evident. Using both detection and attribution and the storyline approach can help to present an overall picture that best represents our understanding of the link between anthropogenically-driven climate change and extreme weather events.

Links:
Severe Weather Event Attribution: Why values won’t go away – Winsberg, Oreskes & Lloyd, Phil Sci Archive, 2019.
Extreme events and anthropogenic emissions – post I wrote a while ago that – I think – makes a similar point to the point being made in Eric’s paper.

Estragon and the Expert

An abstract stakeholder’s dialog. Vladimir, or V, is the expert. Estragon is E.

[V] You have cancer.
[E] OK.
[V] …
[E] Is it curable?
[V] Yes, I guess.
[E] …
[V] …
[E] How?
[V] There’s A or B.
[E] What would you suggest.
[V] If you do A [inaudible], if you do B [inaudible].
[E] Sure, but what would you do.
[V] I honestly can’t tell, I’m just a broker.

---

[E] Well, if you can’t tell and you’re the expert, who can.
[V] You can ask Gwyneth.
[E] I’d rather not.
[V] Well, there’s this study, with some non-suboptimal paths.
[E] I can’t read charts – what do they tell you.
[V] Depends what you want to optimize.
[E] Not dying would help.

---

[V] Look, I’m in the truth-telling business.
[E] I thought you were a doc.
[V] And?
[E] Well, isn’t there an oath or something.
[V] Sure, but I really can’t tell – it would compromise my scientific reputation.
[E] Thanks for setting your priorities straight, doc.
[V] No problem.

---

[E] Alright, doc – do you care if I survive.
[V] Sure.
[E] Good. How.
[V] Do A or B.
[E] You mean I should do A or B, not C.
[V] Or nothing.
[E] You suggest I do something.
[V] Exactly.
[E] At least there’s that.
[V] You can’t say I fell into advocacy.
[E] How about doing A or B.

---

[V] When I say “do A or B,” I’m not advocating for anything.
[E] You’re still helping me not die.
[V] That’s my job – to dispense the best evidence.
[E] Wait. How is helping me not die an observation.
[V] I see. What I mean is that my job is evidence-based.
[E] Advocacy too is.

---

[V] You’re right – I should be more objective.
[E] What do you mean.
[V] I forgot about D, which is only palliative.
[E] Not sure I understand.
[V] A or B will cause suffering, D will alleviate suffering.
[E] Oh, I don’t want to suffer.
[V] You may die faster.
[E] I don’t want to die either.
[V] It’s suffering or death I’m afraid.
[E] There’s still suffering and death.
[V] …
[E] …
[V] I would advocate against that.

---

This stakeholder dialog was meant to convey the idea that experts constrain decisions. These constraints should orient policy making. A broker who’d only describe what’s on the table at very least advocates against what’s beyond it.

Truly brokering preferences does not stop at listing policies – we want to order them in ways to optimize resource allocation along policy paths. Solutions are seldom unique, in which case many experts and many teams may provide more robust ideas. But to compare policy analysis to building a travel site of policies is like comparing pure scientific research to stamp collection.

My point can be reduced to the following Moorean sentence, the paradigmatic example being “it is raining but I don’t believe it is raining”:

[MB] Solutions A or B are most efficient but I’m not advocating for solution A or B.

This looks odd to me. I prefer to think that brokers can be honest and still own their analysis. One does not simply present one’s results and not advocate for them. Scientific advocacy is never arbitrary to the point of being comparable to the advocacy coming from the Contrarian Matrix. In any event, forbiding experts from advocating in their own areas of expertise would be silly.

I recently tried to talk about Moorean sentences. It did not go unpunished. So instead of compacting my point in a single sentence, I made it more explicit by using what I would call an abstract stakeholder dialog. (Patent pending.) The V&E scenario above omits important parameters. Doing something about AGW is a collective decision – V could have cancer too. Both should decide on the cure, e.g. for insurance purposes. The impact of the decision should be made under partial ignorance. as we do not know exactly who will die.

We covered all this innumerous times. While disagreement about details remains possible, I suspect that our ClimateBall episodes furiously recurse because they hit something like dialogical fixed points. Making these points explicit might be more efficient than artificial examples like Moorean sentences. We may need a mix of illustrations and explanations.

We could go on an on with abstract stakeholder dialogs between Vladimir and Estragon and retrace all of ClimateBall. Other characters could be added. Waiting for Godot also casts Pozzo, Lucky, Boy, Godot. One day bots might recreate abstract dialogues for us:

This is amazing. Neural nets trained on subreddits are arguing *with each other* on a subreddit. It's like a tornado made out of tiny tornadoes. https://t.co/pO68i6dzWQ

— Regina Rini (@rinireg) June 6, 2019

The tears of the world are a constant quantity. For each one who begins to weep somewhere else another stops. The same is true of ClimateBall.

A human extinction denier?

Mike Hulme has a new essay that some are promoting on Twitter. He suggests that he is a human extinction denier and objects to the climate emergency narrative. Although I have my own concerns about some of the extreme rhetoric, I found his esssay rather irritating.

Firstly, I think it largely mis-represents the position that he’s criticising. Existential doesn’t necessarily imply the end of the human species. It can refer to other species, some of which seem clearly to be at risk, or to our global civilisation, which some do indeed think is inconsistent with substantial warming.

I also think it falls foul of deficit-model, or linear, thinking. It seems to suggest that there is some logical way to address this issue and that those who promote the existential, or climate emergency, narrative are failing to understand this. I do wish that we could carefully assess all complex situations and then make sensible decisions, but it’s my understanding that this isn’t a realistic way to develop policy. Many factors can influence how we do so, including people promoting extreme narratives.

The essay also highlights that climate change raises a host of ethical, historical and cultural questions that are at most tangentially connected to any scientific findings. This is certainly true, but what it seems to fail to acknowledge is that those he’s criticising may have indeed considered this and that it is these ethical, historical, and cultural issues that have driven them to the position they now hold. People are, of course, free to disagree with their judgements, but why not simply say so, rather than writing something that suggests that your criticism is based on some kind of intellectual authority?

I also think the article produces some rather confused descriptions of our scientific understanding (which is a little odd, given the credentials of the author). For example, it says

Climate prediction science is fundamentally based on probabilistic forecasts which underpin the quantification of risk. There is a range of possible values for future global warming. It is as false scientifically to say that the climate future will be catastrophic as it is to say with certainty that it will be merely lukewarm.

The dominant factor that will determine how much we warm, and hence the impact of the resulting climate change, is how much we emit. I agree that we can’t say that the climate future will be catastrophic, but the more we emit, the more likely it becomes that our climate future will reasonably be described as catastrophic. In a sense this is the key issue; should we limit our emissions in order to reduce the chance that our climate future will be catastrophic?

To suggest that the outcome is simply part of a probabilistic forecast seems to completely ignore that the outcome largely depends on what we choose to do. Much of the climate emergency narrative is motivated by a desire to influence our choices so that we limit our emissions and, hence, ultimately reduce the impact of climate change. Mike Hulme’s essay seemed to completely ignore this pretty basic issue.

I sometimes get the sense that some view this as essentially a binary situation; a world with climate change compared to a world without climate change. It isn’t. It’s a world with a climate that will continue to change while we continue to emit CO₂ into the atmosphere. How much it changes depends largely on what we choose to do. Clearly there are many factors that will, and should, influence how we respond to this. The basics are, however, pretty simple – how much we emit will largely determine how much the climate will change and, consequently, will determine the impacts that we, and future generations, will have to learn to deal with.

Even though there are potentially serious consequences to exaggerating the risks and promoting an extreme narrative, there are also potentially serious consequences to delaying addressing climate change. I think there are valid criticisms of the existential/climate emergency narrative, but I didn’t really find Mike Hulme’s critique particularly compelling. Maybe someone can convince me that there’s more depth to it than it seems.

Links:
Am I a denier, a humen extinction denier – Mike Hulme’s essay.
Existential threat? – post I wrote about climate change being an existential threat.
The benefits of acting now, rather than later – post I wrote about the benefits of reducing emissions sooner, rather than later.

Promises and perils of the Paris agreement?

A few people were sharing, on Twitter, a Science article called Promises and Perils of the Paris Agreement. It mostly seemed unobjectionable. It discusses the linear relationship between warming and emissions, and how this allows one to define carbon budgets (i.e., how much can we emit to have some chance of staying below some level of warming). It discusses Carbon Dioxide Removal (CDR) and how this can’t contribute substantially to limiting global warming over the next several decades. It also discusses the Nationally Determined Contributions (NDCs) that came out of the Paris agreement.

I was slightly confused about the framing, since it seemed to suggest that the NDCs were an alternative to temperature targets. Although they’re clearly based on individual countries determining what they’re willing to do, they do still seem to be linked to the temperature targets and the associated carbon budgets. However, what confused me the most about the article was how it concluded. It suggests that

if societies do end up living in a world in which global warming far exceeds 2^oC by 2100 — which is no longer unlikely, independent of what approach is taken — then it would be far better to do so with a functioning set of democratic global institutions, rather than clinging to fantasies about centralized, detached steering leading to sweeping global transformations, despite decades of experience providing evidence of the implausibility of such an approach.

I couldn’t quite see how this followed from what was presented in the rest of the article. Is it really suggesting that temperature targets are based on centralized, detached steering? Is it something else? I’m not aware of anyone suggesting that we should address global warming by destroying our democratic global institutions and relying on centralized, detached steering, but maybe someone can highlight an example of such a suggestion.

I may be wrong, but my impression is that some see temperature targets as an attempt to define a goal that we should achieve at any cost. Some may indeed suggest this, but I don’t think that this is really the intent. It’s more to do with a combination of climatic tipping points becoming more likely the more we warm, and the impacts increasing non-linearly with increasing temperature. Hence, there is merit in trying to limit warming and – consequently – defining some kind of target. We don’t need to achieve these targets, but the more we miss them by, the more severe the impacts are likely to be.

This reminds me of a quote – from Kevin Anderson – about carbon budgets and the Paris agreement. It seems apt, so I’ll end with it:

It is important to note that the nature of the carbon budgets mean that any failure to deliver deep mitigation rates in the near term (from a high starting value) very rapidly increases future rates to completely unattainable levels. Delay is not an option and our 2^oC mitigation analysis needs to be informed by this.

We can of course throw our hands in the air and declare the implications of such emission constraints are too onerous for us high-emitters to contemplate. But then we need to be honest and say to our and others’ children, as well as many millions living in poor and climate-vulnerable communities, that we have chosen to renege on the Paris commitments. This is an authentic position, allowing others to consider the implications and make whatever contingencies they can to deal with the chaos of the 3-5^oC of warming we’ve decided to bestow upon them.

J R Oppenheimer Asks: Can Science Provide Better Models for Democracy?

I came across a collection of 1950’s vintage essays by Robert Oppenheimer that’s been lurking quietly unattended in my book collection for who knows how long.

Reading essays by scientifically accomplished people from long ago is always an eye-opener. Bits discovered therein have been leaking into my tweet stream this week.

Here is what I take to be the central theme of the essays in this book:

What can and should society learn from an abstract understanding of the practice of science?

We are interested here not in the content of science, but in its process.

Oppenheimer believed that the answer was bigger than “nothing much”. I do too.

 

===

Somewhere in the usenet archives are a few thoughts of my own on the subject from the early 1990s. To my recollection, it was kicked off by ruminations about a TV news program I had seem, entitled (if memory serves) something like “This Week In Science”.

Leaving aside the implausibility of any television network attempting such a thing today, it must be admitted that the program was a failure, a crashing bore. This is because in any given week *Nothing Happens In Science*. Or, more precisely, if something worthy of note happened in science in a given week, nobody will really know that in twenty years.

Contrast this with the sturm und drang of the more conventional news. Personalities, postures, bombastic claims, indictments, excitements, threats of war, hopes for peace! There’s never any shortage of things to talk about,

Ah but come back those twenty years later, and who has made progress? The scientific landscape is utterly altered, while the political landscape is the same mess, with some of the same people strutting on the stage with the same postures?

So I surmised that there is something in the social nature of scientific inquiry that *makes progress possible* which is absent in the social nature of political pursuits (with economic pursuits perhaps holding an intermediate position). And I expressed a hope that everybody else could pick up those skills.

Which skills would they be? A capacity to collaborate with an opponent. A capacity to refine opinions and even reverse them in the face of evidence. A hunger for more evidence against which to test one’s ideas. A genuine pleasure in testing oneself and one’s beliefs.

Science is an odd combination of cooperation and competition. Science is arranged around truth. Yes, if you find the right truths and promote them right, you’ll do fairly well in your career, but everyone knows (or at least should know) that it’s the truths that matter, not your career.

The rest of society is arranged around winners and losers. And it makes little progress.

Do we model ourselves on the steady progress of science, or on courtroom dramas?

It’s clear how the press processes things. This versus that. Opinions from “both sides”. They can’t help themselves. And this is what the public thinks science is about. Warring camps, multiple non-overlapping clusters of opinion. Like pro-choice vs pro-life. Even the nomenclature used by the camps can never overlap.

So my modest proposal was that people adopt what at that time was known without much complexity as a “skeptical” attitude. To spend as much time trying to let go of your own ideas as to convince others of their validity. To remember that the person easiest for you to fool is yourself, and to guard, vigilantly and constantly, against it. To go out of your way to concede those parts of your opponent’s argument as you can bring yourself to do, even if that’s often met with bad faith.

I hoped for seeing, in my lifetime, increasing the zone of consensus among the general public. That’s worked out well so far, hasn’t it?

===

Here is how Oppenheimer frames the question in his 1947 lecture “Physics in the Contemporary World”

( paywalled at https://doi.org/10.1080/00963402.1948.11460172 ; selected excerpts )

“This is the question of whether there are elements in the way of life of the scientist which need not be restricted to the professional, and which have hope in them for bringing dignity and courage and serenity to other men. Science is not all of the life of reason; it is a part of it. As such, what can it mean to man?”

…

“In this field quite ordinary men, using what are in the last analysis only the tools which are generally available in our society, manage to unfold for themselves and all others who wish to learn, the rich story of one aspect of the physical world, and of man’s experience. We learnt to throw away those instruments of action and those modes of description which are not appropriate to the reality we are trying to discern, and in this most painful discipline find ourselves modest before the world

“The question which is so much in our mind is whether a comparable experience, a comparable discipline, a comparable community of interest, can in any way be available to mankind at large.”

…

“Clearly, if we raise at all this question that I have raised, it must be in the hope that there are other areas of human experience that may be discovered or invented or cultivated, and to which the qualities which distinguish scientific life may be congenial and appropriate.”

…

“In the first instance the work of science is co-operative; a scientist takes his colleagues as judges, competitors and collaborators. That does not mean, of course, that he loves his colleagues; but it gives him a way of living with them which would not be without its use in the contemporary world.”

…

“Science is novelty and change. When it closes it dies. These qualities constitute a way of life which of course does not make wise men from foolish, or good men from wicked, but which has its beauty and which seems singularly suited to man’s estate on earth.”

…

“We become fully aware of the need for caution if we look for a moment at what are called the social problems of the day and try to think what one could mean by approaching them in the scientific spirit… In short, almost all the preconditions of scientific activity are missing… All that we have from science in facing such great questions is a memory of our professional life, which makes us somewhat skeptical of other people’s assertions, somewhat critical of enthusiasms so difficult to define and to control.”

…

Oppenheimer closes with “I have had to leave this essential question unanswered: I am not at all proud of that.”

====

I should point out that these excerpts come from a rather long and meandering essay –  this is the unifying theme, but there is much more of considerable interest, some directly and some rather frustratingly tangentially related.

Though it’s a topic I once thought about a lot, I’m not sure I have much to add except that I’d like to hear others’ opinion about it. To motivate the conversation I’ll quote from another essay in the same volume, The Encouragement of Science (1950):

“We need to recognize the situation as new; we need to come to it with something of the same spirit as the scientist’s when he has conducted an experiment and finds that the results are totally other than those he had anticipated.”

As I see things, the complexity of our collective circumstances has been increasing, and the competence of our collective reasoning and planning has not been keeping pace. In the last few years, to the contrary, it seems to have taken a sharp turn for the worse. I don’t see extant political configurations as capable of rising to the occasion. But this will not make the occasion go away. It’s definitely time for some social creativity. Can looking toward successful institutions, the best moments in scientific history not least among them, at least serve as a source of inspiration?

===

-Michael Tobis (mt)

Models are failed hypotheses!

John Christy has written a report for the Global Warming Policy Foundation called climate models have been predicting too much warming. The basic conclusion of the report is that climate models predict far more warming in the tropical troposphere than is observed; essentially it’s suggesting that there is a missing tropical tropospheric hot spot.

One thing that the report doesn’t make clear is that amplified tropical tropospheric warming is not simply a signature of greenhouse gas-driven warming; it’s expected for any kind of warming. I think it’s also worth reading this Climate Dialogue discussion, in particular that by Steve Sherwood, who pointed out that

[w]eaker upper-tropospheric warming and hence weaker water-vapour feedback actually implies, on average, slightly stronger overall positive feedback due to lapse rate and water vapour combined

There are also papers that do indicate amplified warming in the tropical troposphere. I disussed one in this post and another is discussed in this article.

However, what I wanted to highlight was the end of the report, which discusses what one might conclude from the comparison between the observations and the models and which suggests that one option is that

• [t]he models are failed hypotheses.

The report then finishes with

I predict that the ‘failed hypothesis’ option will not be chosen. Unfortunately, that’s exactly what you should do when you follow the scientific method.

The reason that it would not be chosen is because it doesn’t really make sense. Unless a model is so simple that it really only incorporates one bit of physics, it isn’t a hypothesis. Models are typically a combination of multiple bits of physics that, together, allow one to investigate a complex system. Typically, the underlying physics is so well tested that even if the model doesn’t match the observations, one wouldn’t conclude that some of the underlying physics had been falsified.

Of course, a model can be wrong, but that’s not the same as it being a falsified hypothesis. The problem, though, is that all models are wrong, so how does one decide if it’s so wrong that it has no use whatsoever? Even though it might appear that the models are predicting more tropospheric warming than is observed, there are many areas where models have been shown to be skillful.

Determining tropospheric temperature trends is also very tricky, and these are observations that have been corrected on a number of occasions. One really can’t rule out that the mismatch is a problem with the observations, rather than with the models.

As George Box said all models are wrong, but some are useful. Climate models are very useful for trying to infer what might happen if we continue to dump greenhouse gases into the atmosphere. This is especially true if we are interested in trying to understand how it will likely respond on various time, and spatial, scales. If we throw out complex climate models, we’d simply become less well informed. This is not, however, to suggest that we shouldn’t be open about their limitations.

However, our basic understanding of climate change is not really dependent on complex climate models. Our understanding of how our climate responds to radiative perturbations is based on many different lines of evidence. We have models of a variety of complexities. We have recent observations of surface warming, ocean warming, sea level rise, ice sheet mass loss, and many other indicators. We can study changes to our climate that have occured in the past to also understand how it responds to external perturbations. It’s bizarre to suggest that a potential mismatch between complex climate models and observations of one region of the climate system indicates that these models are failed hypotheses.

Even though I’ve been commenting on this kind of thing for a reasonable amount of time now, I still find myself surprised that supposedly serious people will present such simplistic arguments. It’s almost as if they’re looking for something that suits their preferred narrartive, rather than actually trying to help improve our understanding of how our climate is likely to respond to increasing atmospheric CO₂ concentrations. Surely not, though?

Links:
The missing tropical hot spot – Climate Dialogue discussion about the supposedly missing tropical hot spot.
Tropospheric hot spot – a post of mine about a paper that finds amplified warming in the tropical troposphere.
One satellite data set is underestimating global warming – John Abraham article about another paper that finds amplified warming in the tropical troposphere.
Climate change is real and important – article a group of us wrote that also discusses why climate models are actually skillful.
More errors identified in contrarian climate scientists’ temperature estimates – John Abraham article highlighting the occasions on which John Christy’s satellite temperature dataset has had to be adjusted.
Climate model projections compared to observations – Realclimate post showing various model-observation comparisons.

The Carbon Cycle

Robert Rohde, who is lead scientist for Berkeley Earth, has created a really nice illustration of the carbon cycle. It shows how the CO₂ cycles between the different carbon reservoirs, and how our emissions have perturbed the carbon cycle so that atmospheric CO₂ concentrations have risen from around 280ppm in the mid-1800s to over 410 ppm today.

If we were to cease emitting CO₂ into the atmosphere, then CO₂ would continue to be taken up by the natural sinks, and the atmospheric concentration would actually drop. However, as I discuss in this post, there is a limit to how much can be taken up by the natural sinks. As a consequence, between 20% and 30% of our emissions will remain in the atmosphere for thousands of years.

As Robert Rohde pointed out on Twitter, the process that ultimately draws down atmospheric CO₂ is sedimentation, which is very slow. It will probably take more than 100000 years for atmospheric CO₂ to return to pre-industrial levels. We’ve essentially perturbed the carbon cycle so that atmospheric CO₂ concentrations will remain elevated for a very long time.

Credit: Robert Rohde