3.2: Albedo modification

Referring to section 1.1 of this series (or to any introductory course on climate science), one possibility is to try to increase the A term in the radiation balance, i.e., the albedo, to prevent some incoming solar radiation from making it to the ground.

How much would one theoretically want to change the albedo, to achieve a given temperature reduction? The standard number, used in many geo-engineering papers, corresponds toWhich we must of course fight to avoid facing in the future by other urgent means lowering the incoming solar radiation by about 1.8%, with a cooling effect of about 2K relative to a doubled-from-preindustrial-CO2 planet configuration. We can readily calculate this using the simple formulas covered in my first climate post. In particular, the impact of a 1.8% decrease in solar flux in that scenario would be an average temperature of about 285K instead of 286.3K, or a 1.6K cooling.

One could consider modifying the albedo on areas of the land surface, e.g., with mirrors in the desert or reflective paint on roads and roofs and so forth, but only a tiny fraction of the overall albedo comes from the land surface, so increasing that term by a few percent gives a small contribution overall. Thus the field has focused on modifying the atmospheric albedo rather than the surface albedo. Dyson and Marland pointed out exactly this in the 1970s:

One can talk about painting all roads and parking lots white or floating styrofoam plates on the oceans; since the majority of the earth's reflectivity now occurs in the atmosphere, however, it seems most useful to direct our attention there first. The earth reflects about 28% of the incoming solar radiation: 19% from clouds, 6% from air, dust, and haze, and 3% from the surface (Rotty, 1975). Another 22% of the incoming radiation is absorbed by water, dust, haze, and clouds in the atmosphere, and a fraction of this is reradiated to space without ever interacting at the surface. A small increase in the atmospheric backscatter could have a siza ble impact on the atmospheric energy balance.

Dyson, F. J. and Marland, G.: 1979, Technical Fixes for Climatic Effects of CO2 in Workshop on the Global Effects of Carbon Dioxide from Fossil Fuels

Sam Rodriques points out that this does depend on how much one could actually increase the surface Albedo. If it were only by, say, 5%, then the analysis above holds, as 0.05*0.03 = 0.0015, a fraction of a percent, whereas if we can change the cloud albedo by the same percentage, we get a 6x larger effect. But a 50% increase in the surface reflectivity would be a big effect. If 4.5% rather than 3% of the incoming radiation was reflected by the surface, say, then we’d have a fractional change in incoming solar radiation of 1 – (1-0.295)/(1-0.28) = 2%, similar to what is being proposed for the stratospheric scattering particle approach. Because the surface is so non-reflective now, increasing its net reflectivity by 50% could be done with a relatively small area (so to speak) of mirrors.

Now, mirrors are presumably much cheaper to build out over large areas than solar panels, so we can do a fanciful comparison. 18 TerraWatts global energy use / (optimistic 30 Watt per square meter solar panels) works out to an area of solar cells about the size of France. Given a typical surface reflectivity of say 0.1, if we crudely estimate the effect of covering that same area with 95% reflective mirrors instead, we get only a 3.5% change in surface albedo, not nearly enough. If we cover a 20x larger area, that’s a 70% increase in surface Albedo, but covers the whole of China or of the Sahara desert. From a cost perspective, mirrors are so much cheaper than solar cells that they could in a fanciful sense be “cost effective” despite the larger total area needed for mirrors, but doing this would not to be feasible in practice for reasons like local weather effects not to mention issues of land ownership. Mirrors over arctic sea ice, however, have been seriously suggested, as has local modification of the surface albedo via, for instance, more reflective crops, which might also have other benefits. Another fascinating potential land idea is radiative cooling to space!

The seminal early paper in this area involves one of my favorite people, the polymathic and benign Lowell Wood, and offers a taxonomy of ways to do this:
  1. Particles in the upper atmosphere which would back-scatter some of the incoming sunlight
  2. Particles in near-Earth orbit which would back-scatter some of the incoming sunlight
  3. A structure in space between the Earth and sun which would deflect a fraction of the incoming sunlight through a small angle such that it would miss the Earth

Aerosols in the stratosphere

David Keith has an excellent talk that covers other aspects, including potential local effects on precipitation and temperature across the world, early stage ideas about how to properly engineer the particles to maintain the right sizes, lifetimes and distributions, the need for microphysics experiments to constrain the particle design, and many important policy related aspects.The authors, and the field subsequently, seem to have focused heavily on #1, for reasons that they outline:

“Of the three deployments, the stratospheric location is by far the least expensive on a pound-for-pound basis; positioning mass in the stratosphere currently is at least 10^4 times less costly than putting it into low Earth orbit. Moreover, the mid-stratospheric residence time of sub-microscopic scattering particles of anthropogenic and natural origins is comparable to the half decade residence time of its molecular components, so that appropriately fine-scale particulate loadings of the middle stratosphere will persist for five-year intervals.

Arguably none of the other approaches discussed here are within 2-orders-of-magnitude of the economic cost of the aerosol methodIn other words, the particles in the stratosphere approach is cheap, and it naturally needs to be replenished on a timescale of less than a decade, so it naturally disposes of itself if you want to remove it on that timescale. One thing I also wouldn't have guessed is that while this might seem simple we actually don't have the technology to do this currently. It would require planes that are both able to hold reasonable amounts of cargo, our particulate matter we intend to scatter, and be able to reach very high altitudes. Currently no plane can achieve both of these demands.

There is also a precedent for this approach, namely the eruption of Mount Pinatubo in the Philippines (1991), which reduced the average temperature of the Earth by about half a degree Kelvin for a couple of years:

temperature after the Mount Pinatubo erruption

The idea is to recreate the atmospheric effect of such an event a few times a decade.

¹ However it is more uniform than other altitudes and so has that advantage when trying to model effects.Some have proposed putting the particles in the high troposphere where they would, assuming they stay there, not have as much opportunity to do much damage.Unfortunately, as the authors note, the chemistry of the upper atmosphere is complex¹, and in addition to potentially damaging the functionality of the injected light-scattering particles, a worry is that, depending on the material composition of the particles, they could cause some damage to the ozone layer. There is also the risk that putting material into the stratosphere might actually increase heating at this altitude which could change circulation patterns in it.

Recently, Keith at Harvard and other researchers have been studying some of the chemically most benign possibilities, namely using tiny particles of calcium carbonate, better known as “chalk” or calcite. This was highlighted by Danny Hillis in his TED talk, where he emphasized that only 10 teragrams a year of chalk (about twice the mass of the great Pyramid of Giza) would need to be injected into the upper atmosphere to undo the warming effects of the CO2 we already have put in. Hillis points out that this is “a handful of chalk in every olympic swimming pool of rain”, and that the volumetric throughput required to deliver it to the atmosphere would be quite tractable,The definition of a “firehose” I found online lists up to 200kg/sec. This calculation is similar to the one Hillis shows on a literal back of the envelope in his TED talk, in which he explains this firehose analogyequivalent to continuously running just one or a few firehoses.

Note that the positive radiative forcing from the CO2 we have now is calculated to be roughly around (5.35 Watt per meter squared) * ln(408/278) = 2 Watt/meter^2, while the reference radiative forcing in the Lawrence paper was RFGref ≈ 0.6 Watt/meter^2.The paper Hillis is referring to with the chalk appears to be this one from PNAS. From their abstract, “A radiative forcing of −1 W⋅m−2, for example, might be achieved with a simultaneous 3.8% increase in column ozone using 2.1 Tg⋅y−1 of 275-nm radius calcite aerosol.”

The PNAS paper uses Mie scattering theory (a method of calculating the amount and direction of light scattering by particles of a size scale on the order of the wavelength of light or a bit larger) to calculate the change in albedo due to the injection of various sizes of aerosol particle into the upper atmosphere, which causes back-scattering of the incoming light. Anyway, as Hillis points out, a few teragrams per year of this stuff is quite achievable. He illustrates that that corresponds to a big fire-hose operating at full blast. The cost could be very low, on the order of a few billion dollars a year.

The Lawrence paper concludes:

“calcite particles 91 are non-toxic, would not cause significant stratospheric heating, and may counteract stratospheric ozone loss, but their microphysics and chemistry under stratospheric conditions are poorly understood"

Some issues

In general, even if the particles perform their function well, and don’t cause ozone layer damage, there are some remaining big issues:

Does not solve ocean acidification, which is a very big issue. (Though, see Ocean Liming in the last post for something that might be done about it, beyond of course limiting the amount of additional CO2 we release.)

Causes changes in precipitation, in addition to temperature. Pinatubo had precipitation effect including lowering of river flow in Ganges and Amazon. A key concern is that it monsoon seasons could be disrupted by this. Keith points out, however, that directly ballparking this based on the effects of the Pinatubo eruption is misleading since transient albedo perturbations like a volcano eruption are expected to have larger impacts on precipitation than sustained perturbations like ongoing geoengineering.

Impacts different areas differently — although a recent study of an idealized model found that “while concerns about the inequality of solar geoengineering impacts are appropriate, the quantitative extent of inequality may be overstated”.

Could cause heating of the stratosphere — although the calcite rather than sulfate approach could be significantly better on this: “…the radiative heating of the lower stratosphere would be roughly 10-fold less than if that same radiative forcing had been produced using sulfate aerosol”

Changes the light scattering by the atmosphere in a way that may impact ecosystems, in particular making the sunlight slightly more diffuse rather than direct. This will have an impact on the “canopy structure” of forests and other ecosystems. Caldeira seemed to indicate in a talk that he thought experiments to study these effects might be worthwhile at some point in the future. Keith indicated in a talk that overall productivity could increase due to this scattering, as the sunlight is now coming in from a (very slightly) broader range of angles and thus is less shaded in its direct path.

Depending on how and when it would be done, it may not fully mitigate sea ice loss.IAGP also has an integrated simulation of solar radiation management combined with carbon capture and emissions reductions. Interestingly, solar radiation management itself causes some decrease in atmospheric CO2 in that situation through effects mediated by the biosphere. IAGP has a nice integrated simulation, in which they simulate effects such as the difficulty of hypothetical future climate engineers in measuring the extent to which solar radiation management is working, if at all.

In the end the simulated world manages to control the loss of sea ice. But in the case of the Antarctic, things appear complicated. While the ice on Greenland is mostly atop land, in Antarctica melting is heavily driven by processes occurring deep under water. As this detailed paper explains, “Solar geoengineering may be more effective at reducing surface melt than a reduction in greenhouse forcing that produces the same global-average temperature response. Studies of natural analogues and model simulations support this conclusion.We will discuss below some more specific localized engineering approaches that might be able to help where solar geoengineering might not cut it. However, changes below the surfaces of the ocean and ice sheets may strongly limit the potential of solar geoengineering to reduce the retreat of marine glaciers… It may be that significant losses from some West Antarctic glaciers are unavoidable by simply returning climate and oceanic driving conditions to the preindustrial conditions and perhaps that even doing so would not be sufficient to arrest the retreat.

Subpolar-focused stratospheric aerosol injection

One potential approach that is less intense than global aerosol injection is to focus the injections in the subpolar regions, i.e the Artic and Antarctic.

These areas face local warming much more intense than the rest of the planet and the consequences of these areas warming can also be much more destabilising such as accelerated sea level rises.

Deploying stratospheric aerosol injections at just the polar regions has a number of benefits. Firstly, the altitude needed for deployment is lower "Because the tropopause is considerably lower at high latitudes, aerosols or their precursors would not need to be lofted as high, reducing the engineering challenges relative to a global deployment".

Instead of the 20km altitude needed for injections closer to the equator you would only need to reach heights of around 13km if you do you injections at the polar regions. This only simplifies things a bit though, in response to the potential question of whether current aircraft could be used the authors say "after consideration, the simple but surprising answer is—only poorly, and therefore, likely not at all.... a purpose-built aircraft would still be warranted for this mission".

A second benefit is that the actual amount of SO2 you need to release is much smaller (roughly 1/3) compared to if you were trying to have global effects. Not only are you using less material but the effects would be more extreme, for the same set or reasons currently contributing to the polar regions facing accelerated local warming.

(Lee et al 2021) estimates that a 12 Tg-SO2/yr spring deployment at 60°N would force a −3.7 °C annualized average surface temperature anomaly in the region north of 60°N. This estimate was made with a background RCP 8.5 °C scenario but should not be strongly dependent upon the specific scenario.

The program is also, potentially, a very cost effective way of reducing polar warming. The authors stress that this could not be a full replacement for actual mitigation and removal strategies but on a per dollar basis it could be very useful at stopping some of the worst damage caused by polar warming.

The costs are mainly comprised of having to develop a new aircraft design and then running a fleet of over 100 of these models. There would also be the cost of the SO2 used and a fuel surcharge of $0.5/gallon is used to approximate a $50/t cost of carbon. The order of magnitude costs for this program come out at $10bn/yr

There are still severe issues with this approach. As mentioned above no current aircraft would be suitable for this operation and so a new design would be needed and this would then need to be manufactured over 100 times to create a large enough fleet. This fleet would do something like ≈175,000 runs a year or roughly equivalent to the air traffic out of Atlanta's Hartsfield-Jackson Airport, the worlds busiest airport by passenger volume.

Both the creation of these planes a the building out of capacity in the northern and southern hemisphere would take considerable time. The paper estimates something like 15 years meaning this is by no means a quick climate fix. Or in their words:

"The design and build-out of both the flight and ground infrastructure would require more than a decade, such that a large subpolar SAI program is not a feasible emergency response to acute climate stress."

The scheme would also need to be balanced between the two hemispheres otherwise "Arctic SAI has been expected to shift the ITCZ southward, with potentially serious implications for the distribution of tropical precipitation (Robock et al [2008], MacCracken et al [2013], Nalam et al [2018], Lee et al [2020]). Balancing the Arctic injection with an Antarctic injection is expected to nearly nullify such a shift (Nalam et al [2018])." Or in other words if the hemispheres don't receive the same amount of injections global rain patterns could shift.

There are also the traditional governance challenges and public opposition to any scheme involving aerosol injection. Public opposition is so strong to these techniques that in 2021 an experiment in Sweden that was just to test flight equipment was cancelled after public objections. So even though <1% of the worlds population would live in the areas SO2 is being released in public opposition would likely still be a massively difficult barrier to overcome.

"subpolar deployment seems unlikely to bypass the awesome governance challenges that would confront any SAI program, though this would seem to be a crucial avenue for subsequent social science research"