Band Aids for Climate Change #1

Stratospheric Aerosol Injection

Hi, it’s been a while. This post took a long time to write and research properly, and it’s important that I get this one correct in particular, because it’s topical and extremely controversial at the time of writing. This is the start of a series of posts I’m going to write on potential technological palliatives for climate change.

I’m kicking things off with probably the most controversial of the proposed geoengineering palliatives: stratospheric aerosol injection (SAI). Briefly put, this is a form of solar radiation modification (SRM, also known as solar radiation management), which is exactly what it sounds like: that is, modifying the intensity or wavelength of solar electromagnetic radiation. There exist many methods by which one can accomplish this, but today I’m going to talk about SAI because it’s the most controversial. It’s also fairly topical since the “make sunsets” company have made a few headlines recently, and they seems to have gained a bit of online attention.

I’m assuming that my audience has, at first, not heard of radiative transfer and is not aware of how aerosols and clouds affect it. If you know about this already then feel free to skip past the first two sections and start at: Mount Pinatubo.

Here Comes the Sun

Radiative transfer, that is (as the name suggests) the transfer of radiation, is essential to understand for this post. The sun emits electromagnetic radiation composed of a number of different wavelengths, which gets absorbed, reflected, refracted, and diffracted by gasses, aerosols, and clouds in the atmosphere. The (average) spectrum of solar radiation at the earths surface looks like a log-normal distribution with some slices cut out which correspond to the absorption bands of various gasses. This radiation is also diffracted, reflected, and refracted by the atmosphere, which (broadly) is what causes sunsets and other optical phenomena. We call this shortwave radiation.

The radiation which is emitted by the earth is of a different wavelength spectrum, and can usually just be modeled as black body radiation, which has a characteristic spectrum depending on the temperature of the body. The important thing to note is that the radiation being emitted by the earth is mostly infra-red (which is why the greenhouse effect is a thing, a greenhouse gas is that which absorbs infra-red wavelengths while allowing visible light to pass). In addition, I will also introduce two more new terms here: “upwelling radiation” and “downwelling radiation”, which is just a neater way of saying “radiation which goes up” and vice-versa respectively.

So, to summarise, radiative transfer in the context of climate change is a combination of shotwave radiation originating from the sun, longwave radiation originating from the earth, and how the atmosphere affects these different wavelengths in different ways.

Now lets have a look at what the IPCC (which represents the most recent scientific consensus) says quantitatively about radiative transfer. When we are quantifying radiative transfer, we use a term called “radiative forcing”, which means either the accumulation or depletion of energy in the earths atmosphere. For example, “positive radiative forcing” means there is more energy going in than coming out, and vice-versa. This is normally broken down into the amount of radiative forcing which different effects, gasses, or phenomena contribute towards a total number. Figure 1 shows a ubiquitous graph which is from the IPCC AR6 WG1 physical science report, which summarises the contributions by different sources to the change in total radiative forcing between 1750 and 2019.

Figure 1. The ubiquitous IPCC AR6 WG1 figure on radiative forcing

Lets break this down. On the x-axis we have radiative forcing, which is measured in watts (joules per second) per square metre. Each bar then represents a radiative forcing contribution from a different gas or phenomena, for example we can see carbon dioxide causes around 2.16 watts per square metre! There are also error bars on each one, which are important to note. These data come from a consolidation of the last 10 years of research in atmospheric science, and different models predict different outcomes. The error bars represent best and worse case scenario predictions from different model datasets, which can all be found here.

Notice that, among the sea of red ink, there exists a single blue bar: aerosols, and the resulting radiative forcing.

Aerosols and Distribution Thereof

Think of an aerosol as a phase of matter; an aerosol is made from small solid or liquid particles, which are suspended in a gas. Aerosols are extremely important to consider in the atmosphere for multiple reasons: they are a major anthropogenic pollutant (from a number of sources including wood burning and car exhausts), and have an impact on human health (decent overview paper); they act as ice nuclei or cloud condensation nuclei which means that a single aerosol particle can act as a surface for ice or water to form on (here is a very good book for beginners); and they directly scatter radiation. The latter two points here are important for this post, since they are referred to as the indirect and direct parts of aerosol radiative forcing respectively. Note the aerosol bar in Fig. 2 is divided into two main parts: “aerosol-cloud and aerosol-radiation”. The former means aerosols acting as cloud condensation/ice nuclei, then the clouds themselves affecting radiation, and the latter means aerosol particles directly scattering and absorbing downwelling and upwelling radiation.

It was only recently that we realised the the aerosol radiative effect was negative within a reasonable degree of certainty. Take a look at the graph from the IPCC AR5 WG1 report. I’ve not pasted this graph here because I don’t like to have too much whitespace, but the difference between the AR5 and AR6 aerosol radiative forcing estimations is also shown in Fig. 3c, which is shown in AR6 to illustrate progress. Note here that the bar for aerosol direct effects extends into positive forcing, in addition to negative forcing, which partly results from an uncertainty in how much of an effect black carbon aerosol had on the total aerosol radiative effect. In addition, the error bars on both the direct and indirect effects of aerosol are very broad. I’m showing the evolution of the same graph with time to demonstrate that this is an evolving science, and our predictions are continuously improving.

So we have a series of large warming effects and a couple of small cooling effects. The whole deal with aerosol and cloud based solar radiation modification is to take this little blue bar from aerosol and cloud radiative forcing, and maximise it, thus pushing the total radiative forcing to neutral or negative. Simple right? The breakdown of aerosol radiative forcing from different sources from the IPCC AR6 is shown in Fig. 3; this is a pretty informative figure, since we can see the radiative forcing in terms of temperature change in Fig. 3b. The study finds that sulfur dioxide is probably the most cooling aerosol, when considering both direct and indirect effects (the bar is divided into two colours). The reason for the direct effect is simple: sulfur dioxide particles reflect shortwave radiation more than longwave radiation.

The cooling via aerosol-cloud interactions is slightly more complex. Sulfur dioxide produces an aerosol-cloud interaction wherein cloud condensation nuclei concentration is enhanced via the production of ammonium sulfate. More about this can be found here. The TL;DR is: this aerosol causes lots of smaller droplets to be produced in a cloud, which increases the amount of shortwave radiation being reflected.

Stratospheric aerosol injection is when you deliberately distribute aerosol (usually sulfur dioxide) into the stratosphere, with the aim of causing a net cooling effect.

Figure 2. A breakdown of the aerosol radiative effect from IPCC AR6 https://www.ipcc.ch/report/ar6/wg1/figures/technical-summary/figure-ts-15

The reason why people, like the make sunsets company, are doing this in the stratosphere is to capitalise on the large residence times of aerosol in this region, which can be several years, depending on the masses of the individual aerosol particles. The stratosphere is one huge temperature inversion (something I talked about in this previous post) – due to absorption of UV radiation by ozone gas – which causes a low interchange of aerosol between this layer and others.

Mount Pinatubo

On the 12th of June 1991 mount Pinatubo, Philippines, awoke from its 500-year slumber. This was the second largest volcanic eruption of the century in which the event occurred, and the largest near a populated area. Quick thinking and effective forecasting of the eruption saved several thousand lives, however thousands of people were displaced, and the eruption devastated the local environment.

I’m mentioning this eruption for two reasons:

1. About 20 million tonnes of sulfur dioxide was emitted into the atmosphere

2. Good forecasting and planning led to lots of nice measurements of said sulfur emissions, in addition to other aerosol and up/downwelling radiation. We also have good satellite data, unlike with previous eruptions.

These together mean that the mount Pinatubo eruption gives us a neat case study of what solar radiation modification via the means of sulfate particle emissions could do. This paper presents the results from effectively some of the only measurements of stratospheric aerosol injection ever conducted, albeit accidentally. The paper finds that, over the assessed period of four years, a radiative forcing of around “-2.5 Watts per square metre” [sic] was observed – which is huge compared to the predicted average, shown in Fig. 3.

But what does this mean for the stratospheric aerosol injection idea? Well so far I’m trying to present a case for why people are thinking that this will be a potential solution to (or palliative for) climate change. This Eruption is often discussed as a case study as to why the science for the idea is valid (it still gets discussed in overviews), and the IPCC themselves stated “major volcanic eruptions can, thus, cause a drop in global mean surface temperature of about half a degree Celsius that can last for months and even years” in AR4 (back in 2007). So where’s the downside, surely this is solved no?

Well, it was found later in 2013 that the paper which I just mentioned overestimated the radiative forcing by a factor of two, which casts significant doubt over this research and the validity of this case study. This was also after David Keith published “A Case for Climate Engineering”, which is a commonly discussed piece of literature on this topic, and the book on which the “make sunsets” company base some of its statements. A factor of two turns a significant quantity of sulfur dioxide into an impossible quantity of sulfur dioxide, which would likely be quite toxic to human life when it makes its way down (I will talk more on this in the next section).

When we go back to that blue bar we discussed in Fig. 3, and remember that it is split into the direct and indirect effects. The indirect effects are significantly stronger, in addition to being harder to assess. Injecting aerosol into the stratosphere will not trigger the aerosol cloud interaction described in this paper, which is of paramount importance for the overall cooling effect of sulfur dioxide, because the warm clouds (we call clouds made of liquid droplets “warm” clouds) which this effect happens in do not form anywhere near the stratosphere, and when the aerosol descends into the troposphere it is in a chemical form which will not trigger the interaction. It is difficult to decouple the mechanisms which contributed to the temporary cooling effect of mount Pinatubo, but it caused s 0.14C reduction in temperature globally the following year according to this paper (which is the revised estimate, compared to the 0.5C normally attributed by this paper*), which is almost certainly due to the direct effect on radiative transfer in the stratosphere, because tropospheric aerosol would not stay elevated for that long. In addition to this, a recent nature geoscience paper suggests that most of the cooling from volcanoes comes from increase in cloud density in the troposphere, which further weakens the relevance of SAI when compared to other SRM techniques.

So if it takes as much as 20 million tonnes of sulfur dioxide injected into the stratosphere to cause 0.14C of temporary global cooling, what would happen if we skipped this and just injected it straight into the troposphere, thus exploiting only the warm cloud interaction? Well, there would be no stratospheric inversion layer keeping the aerosol suspended, meaning the residence time would be lower, meaning one would have to substantially increase the amount of aerosol we are adding. The main take-away from this section is that no matter what we do, the amount of aerosol we would need to accomplish a measurable negative radiative forcing has historically been vastly underestimated.

*This is still a very commonly quoted figure by people who work with SAI, hence why I have noted this.

Climate Trolley Problem

The first person* to postulate that the global temperature could be reduced by deliberately releasing sulfur dioxide aerosol was P. J. Crutzen in 2006, in an essay. He points out that anthropogenic sulfur has, in part, acted to counter some of the warming resulting from greenhouse gas emissions, due to the reasons I have mentioned above in the first two sections. Since then, a whole field of research has been opened up into modelling the effects of a potential geoengineering scenario (see the Geoengineering Model Intercomparison Project), and the idea has gained traction in the scientific community. Crutzen also mentions in his essay (which I recommend reading) that anthropogenic sulfur has been responsible for several hundred thousand deaths (using the best data in 2006), and points out a dilemma in policy making with respect to these factors.

The problem is simple: do we flip the switch on the track, and attempt to reduce warming rates with SAI, giving what will be viewed by many industrial sectors as a free pass on pollution and a ticket to continue a fundamentally unsustainable economic model, while exposing people to sulfur dioxide aerosol and its atmospheric byproducts in the process. Or, do we consider other options, thereby leaving the trolley on its course, and possibly expose the earth to runaway climate change and all its adverse effects.

It’s hard to say how many deaths worldwide can be attributed to sulfur emissions, because chronic conditions are difficult to attribute to one particular source. However we do know that about 1.3 million deaths per year can be attributed to pollution; sulfur is a large constituent of anthropogenic pollution; and sulfur, when respirated, can cause a potpourri of pathologies.

This is all, of course, assuming that climate change cannot be stopped in time by decreasing emissions, which is the most sensible method. As any good parent tells their child, offering the easy way out seldom provides the opportunity to learn, and giving various parties the tool to continue reaping the benefits of uncontrolled emissions will do just that. My trolley problem also assumes here that SAI is actually the lesser of two evils, which it might not actually be, given both the weakened case of volcanic eruptions and the potential side effects which I will discuss later in unforeseen consequences.

*The first person to suggest this was apparently Russian climatologist Mikhail Ivanovich Budyko in 1974 according to this website, and there are a few people who cite personal communication in meetings from various parties but it is difficult to actually find evidence for this (the books that the website cites don’t give a reference either), and I’m not sure how formulated this idea would be in the context of climate change given that it was 1974. I’m saying that it was P. J. Crutzen unless someone tells me otherwise with evidence in the comments.

Unforeseen Consequences

Aside from the potpourri of pathologies which sulfur emissions can directly cause to human health, it would probably not surprise you to hear that there are potentially adverse consequences from this. Some of these are economic, and some of these are physical. All of them have inevitable indirect human effects.

The first effect is actually directly triggered by the primary cooling mechanism of sulfate aerosol: its interaction with clouds. Sulfate aerosol increases the number of smaller droplets in clouds and decreases the number of larger droplets, thereby making them more reflective to downwelling shortwave. However, these large droplets are important for the formation of precipitation, which means sulfate aerosol would lead to a global reduction in precipitation when it descends. Reducing rainfall will probably cause droughts in areas where precipitation is sparse to begin with, which is the ultimate human effect. This sort of thing can be countered with contingency plans and modelling of geoengineering scenarios to determine better farmland placement. One may correctly argue that runaway climate change also causes rainfall reduction, but it reaches a point where we have to ask ourselves what we are actually mitigating by doing SAI.

Sulfur aerosol in the stratosphere also interferes with Ozone chemistry, thereby leading to a reduction in Ozone concentration. This one is actually a perfect example to demonstrate how just having a net radiative cooling effect is not the same as climate change mitigation. Ozone is actually a greenhouse gas, but it’s a desirable one because increases in skin cancer rate is such a direct human health risk that the warming is preferable, which is why CFCs were banned. This is actually the best example of effective climate change mitigation policy and implementation we have; the hole in the Ozone layer which was caused by CFC release is pretty much gone now. Nonetheless, Sulfur aerosol – albeit through a different chemical process – also destroys Ozone, which means SAI would probably lead to holes in the Ozone layer since the distribution would likely occur in the stratosphere. The inference from this is skin cancer rates would increase in certain areas, which was originally deemed a risk so unacceptable by a number of governments that it led to a significant change in the industry surrounding CFCs in the 80s and 90s.

What goes up must come down. Respiratory conditions and direct implications for human health which sulfur aerosol has are probably the scariest and most observable effects of SAI. The issue is that when sulfate aerosol descends to ground level, it can be in a variety of forms, some of which are extremely detrimental to lung health.

These previous negative effects have all assumed that one would use sulfur aerosol, since this is the most commonly discussed aerosol and the one which has the greatest cooling effect. If one were to find an aerosol particle which would both cause the cooling cloud interaction, without nasty health effects or Ozone destruction, we would also have another factor to consider: ice nucleation. Ice clouds have the opposite effect to water clouds; ice crystals work pretty much in the same way as a greenhouse gas. We can see in Fig. 1 that contrails, which are made of ice crystals, have a net warming effect, which is small but not negligible. When an aircraft flies close to the stratosphere, this is effectively tropopause aerosol injection, since aircraft engines produce aerosol and water vapor in areas of the atmosphere where cloud formation is modulated by either water vapor or ice nucleating particle content. Sulfur aerosol is often dismissed as never nucleating ice, and sometimes actively suppressing ice crystal formation, but most ice nucleating particle studies are conducted under tropospheric conditions. The honest answer to this is that we have no idea what sulfur does in the stratosphere w.r.t. ice formation (and most aerosol will eventually nucleate ice).

Conclusions

I don’t think that this [SAI] is worth it. Most policy decisions in relation to climate change are trade offs between immediate human effects and long-term human effects. At the lower-stakes end of this, using paper straws is a trade off between inconvenience and plastic pollution. In this case the immediate effects to human health are probably a lot worse than the long term climatic benefit, particularly given the recent nature paper.

I will also say at this point that I think banning research into it is a heavy-handed approach which goes way too far. The EU are considering banning it, which I really don’t think is the way forward. Banning attempts by startups like “make sunsets” to actually do it without thinking of the consequences is surely the way forward, since it should not be up to these people to make trolley problem decisions, and the actual development of the deployment technology is not the bottleneck here by any means. The ICCP (different to the IPCC I know, it’s stupid) have made a reasonable statement which I think is a measured reaction.

This was a lot to take in, and longer than most of my posts; I will summarise the points I have made here in this section.

Explanation of SAI

1. Both the sun and the earth emit radiation, with different spectra. This is called upwelling and downwelling radiation respectively.

2. Stratospheric aerosol injection is a method to reduce the magnitude of the downwelling radiation, where reflective sulfur particles are spread in the stratosphere.

Evidence and Volcanoes

1. The eruption of mount Pinatubo is generally given as a case study because there was lots of measurements and it launched lots of this aerosol into the stratosphere.

2. The initial cooling prediction from this eruption was overestimated by at least a factor of two, which is enough to question the economic viability of this technique.

3. A very recent paper predicts that most of the cooling from eruptions is due to increased cloud cover as opposed to stratospheric aerosol, which further casts doubt over the relevance of volcano case studies.

Unforeseen Consequences

All of the following effects will be magnified, since they scale with the amount of sulfur which needs to be injected into the stratosphere, and this amount is larger than anticipated since cooling from volcanoes was overestimated.

1. There will be an adverse effect on human health from increased sulfur emissions. The magnitude of this is difficult to estimate, but about 1 million people a year die from acute or chronic effects of atmospheric sulfur.

2. There will be a reduction in global precipitation due to how sulfur aerosol interacts with clouds.

3. There will be a reduction in ozone concentration due to how sulfur reacts with ozone in the stratosphere, which feeds further into the human health impacts.

4. Ice crystal nucleation rates may increase, which have the opposite effect and cause positive radiative forcing since they backscatter upwelling radiation.