In this episode, I speak with Dr. Claudia Wieners, climate scientist at the University of Utrecht, to explore one of the most urgent and controversial questions in climate science: could Stratospheric Aerosol Injection (SAI) be used to prevent the collapse of the Atlantic Meridional Overturning Circulation (AMOC)?
Claudia explains how AMOC is driven to collapse by two key forces — ocean warming and freshwater influx from increased rainfall and Greenland melt — and why SAI could theoretically counter both. She breaks down the role of the subpolar gyre and deep convection zones in sustaining AMOC, and why the timing of any SAI deployment is absolutely critical: wait too long, and you don’t just fail to save AMOC — you could trigger a dangerous “double cooling” effect around the North Atlantic.
“Every year that we wait increases the chance of being too late, precisely because we don’t know exactly, or we don’t know at all where there is a threshold. So, it could be this year, it could be next year, it could be the year after, or the year after, and so on.”
Dr Claudia Wieners
We also get into the complex trade-offs that make SAI so politically charged: its potential impact on the Amazon rainforest, the risk of it being used as a pretext to slow down emissions reductions, and why many EU policymakers refuse to engage with the topic at all — even though, as Claudia argues, that refusal itself carries serious risks.
Claudia also outlines what a credible research programme would look like: better climate modelling, small-scale stratospheric measurements, and international monitoring infrastructure. And she issues a sobering warning that every year of delay increases the chance we’ve already crossed a critical threshold for AMOC — without even knowing it.
Claudia Weiners AMOC SAI – Interview Transcript
[Nick Breeze – Genn climate]
Claudia, it’s great to see you. Thank you very much for taking the time. Can you start by telling us how the stratospheric aerosol injection geoengineering would interact with the AMOC in order to prevent it, hopefully, from stopping?
[Dr Claudia Wieners]
Yeah, okay. So there are two causes or drivers that can push AMOC into collapse. The one is that it’s not cold enough in winter anymore to make the water so dense that it can sink down.
And the second is freshwater input that can either be an increase in rainfall into the ocean compared to a decrease in evaporation. So there’s more natural freshwater raining in than is going out from evaporation. Or it could be meltwater that could come from greenland melting or from shifting in the CIS distribution.
And now the first bit, the warming, you can directly tackle with cooling the planet, obviously, including from stratospheric aerosol injection. And also, in general, stratospheric aerosol injection decreases the intensity of the water cycle. So in global warming regions which have already a lot of rainfall get more of it, and regions that have a lot of evaporation get more of it.
And the SAI turns that back, so it will probably also reduce the nettle rainfall into this part of the ocean. And finally, cooling helps against meltwater input. So it could reduce or remove the extra input of meltwater from Greenland.
And so if you do it in time, so AMOC is not yet dying, but you remove the things that are trying to push it into a dead state, then you may prevent the AMOC tipping.
[Nick Breeze – Genn climate]
Okay, I want to come back to the time factor in a moment. But when you gave a talk in Cambridge, which I watched, there was also a reference to the subpolar gyre. Can you talk a little bit about how these two interact?
Because they seem to be both sort of vital parts of the story in a way.
[Dr Claudia Wieners]
Yes. So the typical picture of AMOC is water moves northwards, then it becomes dense because it loses heat to the atmosphere, which is why we have a nice weather in Europe, then it sinks down, and then it returns on the bottom. But we don’t take into account east-west direction, right?
It comes in from the south, and then it goes back to the south. So now, if you look into more detail, these waters don’t go straight north and then south again, but they curve a bit around. So you have the Gulf Stream that comes from the coast of America, it turns to Europe, and then part of it curves around Iceland and around Greenland, so the Labrador Sea, and then it goes back.
So you have a kind of counterclockwise movement, which is partly related to AMOC and partly also to local winds. You do say for AMOC you need water to sink, but it’s not such an easy story, because water can’t just sink somewhere in the open ocean. There’s mathematical reasons for that.
However, if you’re near the coast, then this balance that forbids sinking in the open ocean is broken, because you have extra forces in play, namely the friction against the coast. And together, if you do the math, this means that if you have a current that moves around and fills the coast on the right-hand side, and the water in this current gets denser and denser, for example because it cools, then there can be sinking along the coast. So basically, the sinking of AMOC doesn’t happen anywhere, but it happens when it curves around these coasts.
But then, away from the coast, inside the ocean, there’s a second phenomenon, which is called deep convection or vertical mixing. So with mixing, I don’t mean a large-scale sinking of water, but I mean that here one water parcel moves down and another one comes up, so we have a local vigorous mixing up and down. And this can be triggered by winter cooling.
So we know that cold water is denser than warm water, so in the autumn, the surface layer has warmed up, so it wants to stay up because it’s lighter, but now we get the cold winter air blowing over it. It extracts heat from the ocean in this interior away from the coast, and then individual water parcels get so dense that they want to sink down, and then lower water parcels move up to compensate. This means that eventually the whole water column in the middle of, for example, the Labrador Sea will cool down, because the lower water also has a chance to see the cold winter air.
And that has two consequences. First, it makes it easier for this coastal current to remain strong. And second, there’s an exchange of water from that column with the coastal currents through little whirls, little eddies.
So this water in the interior also helps to densify the current around the coast, and that’s what we need for the sinking.
[Nick Breeze – Genn climate]
Now I want to come back to the the timeline thing, because in the research that you were sharing, you use 2020 as a point to start deploying stratospheric aerosol injection. But obviously 2020’s passed, and we’re in 2026, and it’s a sort of globally chaotic world. We’re not very rational at the moment.
How important is that time frame in terms of where we are now, and what’s likely in the real world?
[Dr Claudia Wieners]
Well, there’s quite a lot of uncertainty, right? Because climate models are not perfect, we don’t know perfectly how much CO2 we’re going to emit in the future either. And we also don’t know perfectly how well SRM works.
So as you just said, the subpolar gyre, this convection thing, and the AMOC help each other, which doesn’t mean they’re equivalent. So it could be that, for example, the subpolar gyre could tip first, and then AMOC could tip a weekend later. And the little bit that we know from models suggests that maybe the subpolar gyre is at risk more quickly.
And if the subpolar gyre tips, this at least increases the risk of AMOC also tipping, because this helping system is then taken out. So there are some climate models which suggest that the subpolar gyre, or parts of it, could collapse relatively quickly. So in the models that we use, it’s very enthusiastic about doing bad things here.
So it’s very enthusiastic about tipping the subpolar gyre. And then actually, even in the 2020 case, one part, the Labrador Sea, was not possible to maintain, even with the early SEI. But the other two parts, so that’s the Irminger Sea between Iceland and Greenland, and then another convection spot somewhere north of Iceland, these you could save with the early SEI, but not if you wait with the SEI until 2080, because once this convection zone is dead, then the surface water becomes too fresh, so too little salty, and then it’s too less dense to start the convection again.
Now, it would be really cool to do some more simulations that we don’t only start the SEI either in 2020, which is past, or 2080, which is a long time hence, but also for intermediate cases. And so far, we unfortunately haven’t done it in our group. But we are now working on looking at additional simulations from other groups that have some intermediate starting date, so we hope we can fill in a bit the gaps.
But so far, we only know that if you wait until the collapse of the subpolar gyre has happened, you’re not going to restore it. If you’re in time, you may save at least part of it.
[Nick Breeze – Genn climate]
Okay, and you said that if you wait to this period, like 2080, it’s not going to work. Can you talk about what would happen if you did deploy it at a later stage once we already see the AMOC collapsing?
[Dr Claudia Wieners]
Well, I mean, so first, I should maybe add a caveat that’s from the model that we use, but models don’t perfectly agree with each other. So I’m not guaranteeing that what I say is fully correct, right? But at least it’s a plausible scenario.
So you can definitely not say just because my model says that collapses in 2052, that’s exactly the correct year, right? But we can at least say, if a collapse has happened, it’s rather unlikely or at least less likely that SEI will be useful. So suppose that we start too late, then we have already collapsed subpolar gyre convection.
And in our simulations, we have also strongly weakened AMOC, so AMOC is not fully dead. But at least it’s weakened substantially. And if now you do the SEI at the moment, that it’s already too late to restore the convection, at least for the next 50 years or so, then because so you get a cooling in Europe and the surroundings of the North Atlantic from the collapse of the deep convection and the weakening of AMOC.
And that’s already strong enough that at least the region around Iceland will be colder than without the global warming. So even if we have a lot of global warming, this region can cool because of this collapse of these ocean currents. Now suppose that on top of that you cool extra because you put SEI in it, which does not restore AMOC, AMOC remains dead, that’s one cooling and you put SEI, that’s another cooling, then you sort of double cool the region, and then you get a really significant cooling, let’s say around the North Atlantic.
And it’s probably not what you want, at least not if you’re in Iceland or Norway or these kind of surroundings.
[Nick Breeze – Genn climate]
Okay, so that’s a critical point to understand in terms of what these models imply.
[Dr Claudia Wieners]
Yes, it could be that, at least from the perspective of, let’s say, AMOC tipping or weakening, you could make the impact worse. If you think that you’re restoring the problem, because if you wait too long and you get this double cooling, you’re worsening impacts. Of course, there’s an important caveat, which is that AMOC isn’t the only problem in the world.
Suppose AMOC hasn’t yet tipped, so there’s a chance that we can save it with SEI. But if we save AMOC, but we make it worse for the Amazon rainforest, that’s a very, very difficult dilemma to handle. The impact of SEI on the Amazon rainforest is pretty uncertain, but we cannot exclude that it could make it worse.
[Nick Breeze – Genn climate]
Can I just ask very quickly, is there a negative impact if the AMOC does collapse on the Amazon?
[Dr Claudia Wieners]
So there’s two things. AMOC collapse will be bad, at least for part of the Amazon, because if AMOC collapse, the Northern Hemisphere cools. If the Northern Hemisphere cools, then the tropical rain band, called the Intertropical Conversion Zone, it will shift southward because the ITCZ, the rain band, likes it warm.
So if you cool one hemisphere, it’s going to move to the other one. And then at least the northern part of the rainforest will receive less rainfall, which could cause the forest to die. Let’s even suppose that AMOC stays constant.
SEI itself does a lot of different conflicting things to the Amazon rainforest. First, it makes it colder, so there’s less heat waves, which is good because heat waves trigger forest fires. But it also causes less precipitation, which is bad because rainforests like rain.
It also causes less evaporation, which is good because the forest likes soil moisture. And we don’t know which of these will win, and it could also depend on how exactly we do with the SEI. And then on top, there’s the thing that AMOC collapse is also bad for the rainforest, and stopping AMOC collapse is then good for the rainforest.
All these effects are intermingling, and we don’t really know what’s going to dominate. But again, it could be that SEI is such not good for the rainforest. It’s currently not excluded.
And now if you think there’s a risk for the rainforest, so do we save AMOC, or do we sacrifice the rainforest? Or maybe at 50% chance, you sacrifice the rainforest. There’s a lot of possible trade-offs here.
We cannot look at just AMOC and think we take our decisions based on that.
[Nick Breeze – Genn climate]
Yeah, it is interesting how, I suppose, because we’re in the eye of the storm of an AMOC collapse, we tend to prioritise it very, very highly. You mentioned Iceland a minute ago, but Iceland has actually declared the AMOC collapse a national security threat. That’s unusual for any state to say something about a tipping point being to recognise it so clearly as a threat.
Has this increased the engagement with policymakers or other researchers or other entities that you’ve had with your research? Because it seems like these conversations we’re not having, and you’ve just mentioned the complexity of the Amazon, for example, we surely need to be trying to unpick some of this stuff and try to figure out.
[Dr Claudia Wieners]
Yes, I do think that in the EU, from my limited sampling, policymakers are, well, sort of getting more aware of SRM, but they mostly don’t like it. And there are a lot of things that you can dislike about SRM, obviously, right? I mean, it’s politically risky, it could distract us from mitigation, because what if Shell tells us you can keep driving your car, it’s no problem because we have this wonderful plan B, so you don’t need to mitigate.
So there’s a lot of things not to like here, but this causes some policymakers to not want to engage with it, because for now, SRM is an icky topic. We don’t want that. We don’t want to be seen near that.
We don’t want to feed conspiracy theorists with ideas that we are going to do that because then they will vote against us. So there’s a lot of cross current. So right now, SRM is still icky, which, by the way, makes it very hard to get research funding.
But the unwillingness to engage, I think, is very mistaken. Because so first, from a physics perspective, suppose SAI is the one thing that stands between us and AMO collapse, and maybe even doesn’t endanger the rainforest. So suppose it could be that SAI is from a pure physics perspective, a very powerful tool to prevent really, really big problems, then you could argue that we cannot afford to dismiss such a powerful tool, even though we acknowledge that it is a dangerous and B, it should not be the first priority, because we anyway need to mitigate like hell whether we do SRM or not.
And then the second thing is, okay, suppose that you refuse to see it in this way. So you really think that SRM is a really bad idea. Okay, you can think of it if you balance the political risks of SRM very highly compared to the potential physical benefits, then it’s a legitimate attitude, even if I disagree, to say we don’t want that.
But even then, if, let’s say, the Netherlands or Germany say we don’t want it, they should be aware that there’s a chance of someone else wanting it, let’s say, within the next 10 years or so. So you can’t prevent someone else from saying we have now been hit with such a terrible impact, or we are so afraid of AMOC or whatever. So we want to push through with that.
So at minimum, we need to then know what’s going to happen here. If someone says we do it, we want to do this and this and this. And then at least if you know that this kind of implementation has really bad consequences.
So if you really have to do it, at least please don’t do it like this, but do it like that. So even if you think we shouldn’t do it, you need to engage with the topic so that you’re prepared. And I have the feeling that policymakers in certain EU countries that I had the pleasure to see, they not all want to understand this point.
[Nick Breeze – Genn climate]
But it seems like we have to reach a sort of reality point where, I mean, no one in their right mind would say, oh, this sounds like such a fabulous idea. But at the same time, the alternatives bring you into sort of focus, if you like. If you were in a point, in a situation where the response changed, and they say the EU said, look, we need to know more.
And we’re going to put a significant budget behind this. And how would you see a research programme that reached out to all the various groups? How would such a thing progress?
And what would be the sort of the early stage, vital steps we would need to know?
[Dr Claudia Wieners]
Um, so I don’t think I can hash out the complete programme, especially on the social science side, on the government’s questions. I mean, and I really acknowledge governments is very important, but it’s not my expertise. From the physics side, I think you can.
So first, as the modelling part, and we will absolutely need climate models, because we can’t just test in real life and collapse and then see what SAI does. I mean, the whole thing is that we want to avoid that. And then there’s the experimental side.
So if you look at the effectiveness of SAI, we know roughly that with sulphate results, it should work because we have had volcanoes to test it for us. But we also didn’t have that many volcano measurements. So there are some amount of uncertainty.
So do we need like one Pinatubo eruption per year or have a Pinatubo eruption per year, two Pinatubo eruptions per year for a certain amount of cooling? Well, we know it’s not 20 and it’s not 0.1, but whether it’s half or two, we don’t know yet. And it depends on how the aerosols will fill from droplet sizes and where you do it and so on and so forth.
And some of it you can model, but models have a certain amount of disagreements when they disagree, like we see in a factor of two or three, and you could narrow it down with more measurements. But if we can’t ask Pinatubo to do a tailored eruption for our measurement needs, we would need to put the stuff there ourselves. And right now, if you propose such an experiment, it will be shut down.
We’ve seen for Scopex, well, there was a lot of complex reasons for that. But it’s very hard to do these kind of things because it’s considered icky. And I think so what would be really helpful here would be that we would have clear rules, like you can not do a big experiment, like something big would at minimum be like it does something to the climate.
So that’s big. So for now, we don’t want that. It’s too risky.
But you can do a small experiment that is so tiny that it doesn’t really have any environmental input. Suppose I would say, well, I have nothing to do with SAI, but I want to research aerosol distributions in case Pinatubo erupts again, then nobody would think twice about it, right? If you do a small amount of research, it’s too small to have any bad environmental input because it’s just impact.
It’s just a few kilogrammes of stuff. They would say, sure, Pinatubo could erupt again. So let’s research what’s going to happen in this case.
If you say it’s for SAI, they will say, oh, no, no, no, no, slippery slope. You can’t do that because then you go to the bigger and bigger stuff. So maybe we should say what you can do for Pinatubo, you can, like everything that’s sort of small, you can do for SAI, but everything that’s big enough to have impact in itself, we can’t do for now.
And maybe if in 10 years time or five years time or whatever, knowledge has progressed and we’ve seen that SAI seems a better idea than we thought previously, then we could open up and come up with other experiments. To narrow down model uncertainties, we will eventually need small experiments. And we need to know what the stratosphere looks like without anything.
And of course, we do know quite a bit, but a first step that could make sense from several point of uses, invest into aircraft that can measure the stratosphere even without experiments. So this could be used for experiments maybe later on. They could be used to establish the baseline state without anything like any eruptions or any SM going on.
And they could also be used as a kind of political safety tool, because suppose that there is some kind of conspiracy going on that someone claims that China is already doing it. Now suppose that India and Europe both have aircraft that can measure aerosol concentrations and both of them were not necessarily friends of China, but also not maybe friends of each other. Both of them confer, well, we haven’t seen anything to independent sources.
Maybe the US also has an aeroplane that could take out some conspiracy theories. So you could have people watch each other and be sure that nobody’s doing crazy stuff. So investing in observational gear could also be a wise step, even if you don’t want to do experiments yet.
And then there’s the modelling side. So one long-term issue is climate models have problems with simulating some tipping points. So we need to improve models.
We need to do that anyway, regardless of whether we want to consider SRAM or not. But there’s also quite a lot of things that you could do with existing models in the short term to get, for example, a first impression of what happens if you do the SAI in 2050. So you could sort of do my experiments in more climate models with more intermediate time steps.
And we know that we can’t fully trust them, but at least we can get a somewhat better estimate. So we need more modelling, more people dedicatedly looking at what does SAI do to tipping points. If you can’t fully model tipping points, at least what it does to the variables that push tipping points.
We need to improve the models on the tipping side, but also on the SRAM side. And for that you may need the experiments, small-scale ones to better understand things like aerosol distributions.
[Nick Breeze – Genn climate]
I just wanted to ask, I know you said earlier that you haven’t done yet the research for later time increments in terms of…
[Dr Claudia Wieners]
Except 2080, yeah.
[Nick Breeze – Genn climate]
Yeah, except 2080. Is there no indication at the moment of, say, a time frame that we could consider absolutely vital, that we get something, at least, you know, a knowledge base or something in place, like 2030, 2035? Is there a sort of time that we go beyond where it becomes too late to consider this?
[Dr Claudia Wieners]
Probably yes, but it’s very hard to be sure what time that is. So, because for now, unless the model improves very much, we don’t know when the subpolar gyre collapses, or for that matter, when AMOC collapses. We also don’t know if a subpolar gyre collapse guarantees an AMOC collapse later on.
So, if that’s the case, suppose a subpolar gyre collapses early and AMOC, without subpolar gyre, cannot exist. Well, it will continue to exist for a few decades, but it eventually has to die. That would be a really dangerous scenario, because then we could be there any time.
We could even be there now and not knowing it. So, if the subpolar gyre just starts to take 10 years until you’re sure it has died, we could be there now or in a short moment, or we could not be there for another 30 years or longer. So, for now, all we can say is, from a pure physics perspective, I mean, there are very many political reasons why we don’t want SEI right now, but from a pure physics AMOC perspective, which is a very narrow perspective, we could say that every year that we wait increases the chance of being too late, precisely because we don’t know exactly, or we don’t know at all where there is a threshold.
So, it could be this year, it could be next year, it could be the year after, or the year after, and so on. And every year that we wait increases the chance that we have crossed it. So, if you really narrow-mindedly only look at AMOC and nothing else, then you would say the sooner the better.
But politically and otherwise, we cannot do such a narrow-minded viewpoint, of course. What happens to the Amazon? What happens to Shell abusing it rhetorically?
So, we should be humble and accept that SRM is only a part of the puzzle, which is a part of the puzzle. So, SRM is one possible tool against climate impacts and could be powerful, destructively or constructively, but the bigger picture is we need to get rid of the cost. So, that’s the CO2.
So, we need to mitigate this. If you’ve never heard of SRM, as quickly as possible, because every tonne of CO2 that we save will reduce risks, either because we have less global warming or because we need less SAI to quench the global warming. But around that, so climate change is also only part of the package.
Because I think, so we can try to solve climate change in a very technological way. We put solar cells, we put batteries and so on, but probably we’ll still not make it without societal changes. The rich need to become way more modest so that the poor can have a proper life.
And so, I think we also need a huge economic reset. And even if climate change were not an issue at all, I think it’s crazy that the richest person in the world has enough money to give everyone extreme poverty half a year income. So, we should not blindly stare at climate change, but we really need to reset quite a lot of things.
And that’s often overlooked in this debate. And obviously, that’s my opinion.
[Nick Breeze – Genn climate]
We’re not really so much talking about a cure, we’re talking about a sort of intervention, whilst we’ve got all of these other emissions issues, which are completely out of control. And the burden of greenhouse gases is so high at the moment.
[Dr Claudia Wieners]
So, in the Netherlands, you have the saying that prevention is better than healing for diseases. So, from a climate perspective, the prevention is to stop emitting the CO2. But that’s not enough, you also need to heal the disease which is there.
So, that’s the carbon dioxide removal, you take out the illness, the CO2. And then even below that, then you have the SRM, which is symptom fighting. So, that’s the paracetamol against the fever.
But the cause of the fever is something else. It doesn’t mean that SRM is not useful, if the symptom is an extra half a metre of sea level, extra seven metres of sea level or whatever, you may want to symptom fight as well. But we need to get rid of the disease.
[Nick Breeze – Genn climate]
We’re very much, you know, talking about tipping points. And that is this sort of point that’s irreversible on timescales that matter to us.
[Dr Claudia Wieners]
I mean, if you have a patient who is at risk of dying from fever right now, because his fever is 41 degrees Celsius, so that’s, I think that’s sort of the dangerous range that you can die from the fever itself, then you need to give him a paracetamol, because otherwise he’s going to be dead before he is in the hospital for the other treatment, right? So, the fact that SRM is only symptom fighting, it’s not a complete argument against it. So, maybe we need to symptom fighting, but we absolutely need to get rid of the disease.
So, we need to mitigate this if you’ve never heard of SRM. And maybe you should also consider whether or not SRM is a good idea. So, that’s my task, so to say, but it’s a small part of the puzzle.
[Nick Breeze – Genn climate]
Well, look, it’s been great to speak to you. Thank you very much. And it’s very enlightening.
[Dr Claudia Wieners]
Okay. Yeah. Thanks.