Correction: Figure 3 below was originally Figure 3 from the Cao/Caldeira paper instead of the correct Figure 1 from the paper. This has been fixed.
In 1992, the National Academy of Sciences defined “geoengineering” as the “large-scale engineering of our environment in order to combat or counteract the effects of changes in atmospheric chemistry.” The most significant changes in atmospheric chemistry today are the emissions of greenhouse gases into the atmosphere by human activities, especially but not limited to carbon dioxide (CO2). In recent years, climate scientists have begun to investigate whether or not geoengineering is practical as a means to give humanity the time it needs to adapt to climate disruption or, as some would prefer, a means to controlling the environment such that no changes in energy consumption patterns are even necessary.
The results of three different geoengineering studies were recently published, and all three found that geoengineering would be fraught with unintended and unexpected consequences.
The first two studies looked into the effects of pumping gigatonnes of sulfur dioxide (SO2) into the stratosphere. This method of geoengineering works by decreasing the amount of energy that reaches the Earth’s surface by increasing how much light reflects off the upper atmosphere. Scientists know that this would work because the Earth cooled for a few years after the eruption of Mt. Pinatubo blew large amounts of SO2 into the stratosphere in 1991. But what scientists don’t know is what the side effects of using SO2 in a long-term geoengineering project would be.
The first study was performed using two different climate models to project what would happen as a result of pumping 5 Tg per year (5 Mt per year) of SO2 into the stratosphere. The scientists ran the models over several decades and compared the results to a baseline CO2 emissions scenario (A1B) assuming rapid economic growth and spread of new energy technologies, a maximum global population of about 9 billion people, and reduced disparities between wealthy and poor nations.
The simulations showed several things. First, continuous pumping of SO2 into the stratosphere cooled the average global surface temperature in less than a decade to approximately pre-industrial temperatures. Second, injecting the simulated amount of SO2 delayed the increase in global mean temperature approximately 30 years. Third, when the temperature returned to approximately the same level as the simulation starting point, the injected SO2 had significantly altered the global temperature pattern from the 1990-1999 means. For example, the Amazon basin was about 1 K hotter, the Arctic still showed polar amplification due to losing very reflective sea ice, while Australia was actually cooler and sub-Saharan Africa were generally cooler. As the authors pointed out, this means that
increases in GHG concentrations can still have a profound impact on regional climate even if geoengineering is successful in counteracting any change in global-mean temperature.
Fourth, along with changes in average global surface temperature, there are also changes predicted in global precipitation. For example, the simulation projected that much of the US Midwest would dry out along with the Texas and Mexican Gulf coasts and the northern area of South America. On the other hand, inland areas of sub-Saharan Africa would get wetter along with northeastern Australia and the Siberian coast.
The dashed lines in Fig. 1 (Fig 3 from the paper, at right) illustrate the fifth result of the simulations, specifically the response of the average global surface temperature if the SO2 pumping was ever turned off. The dashed line in Fig 1a (HadGEM2 model) shows that the temperature rise after turning of the SO2 is projected to be more than double the highest rate of temperature increase from the A1B scenario itself. While the simulation shown in Fig 1b (ModelE model) behaves somewhat differently, both show a rapid increase in temperature starting immediately after the SO2 pumping is turned off. Given there is concern about whether ecosystems and nations could adapt to the rate of global temperature increase defined in the baseline scenario, doubling that rate should be cause for greater concern.
The authors of the study also called for other climate modelers to attempt to duplicate this investigation’s results using other climate models and, ideally, using a standardized set of test conditions in order to make direct comparisons easier.
While the first study found that average global surface temperatures could be delayed by about 30 years by pumping 5 Tg per year of SO2 into the atmosphere, the second study used a different simulation to determine what would happen if the amount of SO2 was increased continually to keep the average global surface temperature stable.
Like the first study, the second study started from the A1B emissions scenario, but then ran 54 different simulations where differing amounts of SO2 were pumped into the stratosphere in order to counteract and then stabilize the average global surface temperature at various points. In addition, the researcher divided up the world’s land area into different regions in order to estimate what the overall effects of geoengineering would be on each region.
What the researchers found was that there appeared to be a fundamental trade-off involved in using SO2 to stabilize the average global surface temperature. Specifically, the results appeared to show that more stability in global temperature meant less stability in regional precipitation. In addition, the researchers determined an “optimal” climate where the changes in both temperature and precipitation were held as low as possible, and then they mapped those climates into Figure 2 (Fig. 4 from the paper, at left). Fig 2a & b show the optimal climate in the 2020s vs. the 2070s respectively, following 15 or 65 years of non-stop SO2 injection into the stratosphere. The red and orange regions represent those areas where “optimal” is achieved using less SO2 pumping while the blue and dark green regions are those areas where more SO2 pumping is “optimal.”
What Figure 2 shows us is that there’s a number of clear differences between those areas where more SO2 pumping would be preferred vs. those areas where less SO2 is better. For example, the northern hemisphere generally wants more SO2 than the southern hemisphere. But perhaps the most significant is that the developing world, especially western Africa, India, and the island nations of south-east Asia are all going to want less SO2 pumping, while the developed world (the US, Europe, China, Russia, Japan) will all want more SO2 pumping.
This split between north and south, developed vs. developing is very likely to cause conflicts between regions and nations that will complicate any SO2-based geoengineering system. As the researchers point out, these results
suggest that as our understanding improves, serious issues of regionally diverse impacts and inter-regional equity may further complicate what is already a very challenging problem in risk management and governance.
While it’s theoretically possible that more advanced geoengineering technologies than simple SO2 pumping could be tuned to get the regional responses to be closer to “optimal,” no technology current exists or has even been proposed that could, for example, bring the optimal precipitation and temperature response of India more in line with that of China. Furthermore, this research focused only on two climate metrics – temperature and precipitation – while the paper says there are any number of other metrics that could be optimized for, such as not shutting down a monsoon or retaining sea ice. And the regional effects are also not the only issue, as conflicts between nations/regions and industries (like maritime shipping) could also develop over such things as the importance of retaining summer sea in the Arctic.
The results of the two SO2 studies agree broadly with each other – pumping SO2 into the stratosphere will cause unintended changes in precipitation globally but with some regions faring better than others. The second study, however, only ran their simulations until 2070. In reality, unless some method of accelerated CO2 removal was implemented, the geoengineering of the stratosphere with SO2 would have to go on for centuries.
The third study investigated the effects of an ideal system that was able to remove CO2 from the atmosphere to project what effect it would have on average global surface temperature.
The researchers ran three different simulations. The first, baseline simulation followed a high CO2 emissions scenario until 2049 and then, in 2050, dropped the emissions instantly to 0 but did not otherwise reduce the CO2 in the atmosphere. In the second simulation, the emissions not only dropped to 0 but all the excess CO2 already in the atmosphere was also instantly removed, but any extra CO2 later released into the atmosphere by the ocean or the biosphere were not removed. The third simulation also removed any extra CO2 added to the atmosphere by the oceans and biosphere as it was added, simulating continued CO2 removal by some geoengineering technology.
Figure 3 (Figure 1 from the paper, at right) shows the CO2 concentration and the global mean surface temperature as they change over the course of the simulations. The first simulation described above (“zero CO2 emissions”) is the dashed black line, the second simulation (one-time CO2 removal) is the gold line, and the third (“maintenance of pre-industrial CO2“) is the red line.
There are a number of key features of these two graphs. The first key feature is that CO2 concentration doesn’t fall rapidly even after CO2 emissions have been reduced to 0 (top graph, black dashed line). This is because the lifetime of CO2 emitted into the environment is thousands of years, so the CO2 concentrations fall about 100 ppm over the course of 450 years unless the CO2 is actively removed from the atmosphere. The second key feature is that temperature continues to increase even after all CO2 emissions are stopped (bottom graph, dashed black line). This is because the ocean stores massive amounts of energy and it responds relatively slowly. As a result, the global mean surface temperature in 2500 is simulated to be about the same as in 2050.
The third key feature is that even after all the anthropogenic CO2 is removed from the atmosphere in 2050, the CO2 concentration rebounds within a decade or so to around 360 ppm, restoring almost half of the CO2 back into the atmosphere (top graph, gold line). As a result, the effect of the one-time CO2 removal on global mean surface temperature is to only cut the temperature by about half instead of fully back to the pre-industrial level (bottom graph, gold line). The reason this happens is that much anthropogenic CO2 is absorbed by the ocean and the biosphere, and when the atmospheric CO2 concentration drops, the oceanic CO2 comes out of solution and the excess CO2 held in the biosphere gradually re-enters the atmosphere as the fertilization effects of excess CO2 fall.
The fourth key feature is how fast the temperatures respond to CO2 removal. If it were physically possible to remove all the anthropogenic CO2 from the atmosphere, then the global temperature would fall by about a degree Celsius within less than a decade (bottom graph, gold line). And if the excess CO2 being emitted by the oceans and biosphere back into the atmosphere were also removed via some technology as they were released, then the temperature would fall to nearly pre-industrial levels within 70 years (bottom graph, red line).
Beyond the fact that there exists no technology that can instantly remove and sequester hundreds of Gt of CO2 from the free atmosphere, this study points out a number of problems. First, merely transitioning to zero CO2 emissions by capturing CO2 from power plants and switching to renewable energy sources won’t enough. Second, in order to return the global mean surface temperature to about pre-industrial levels, the total amount of CO2 that would need to be sequestered is almost equivalent to the total amount of CO2 that was emitted in the first place. That’s a LOT of CO2, and so CO2 removal will be a long-term and difficult process.
Combined, these three studies paint a bleak picture of geoengineering that runs counter to some proponents’ claims. Pumping SO2 into the stratosphere can generally delay or counteract the increase in average global surface temperature, but at the cost potentially serious changes in precipitation and the accompanying changes in various regions’ ability to sustain ecosystems and civilization. There will be other unintended and unexpected consequences from using SO2 to geoengineer the climate, and given the inability of international bodies to presently manage conflicts over climate disruption, it’s unlikely that those same bodies will be able to manage the regional problems that will occur as a result of geoengineering. While more research into better models, the effects of SO2 pumping on sea ice and monsoons, and more effort on standardization of models and tests are all warranted, there are some issues that no amount of geoengineering with SO2 can impact, such as the fact that only reducing CO2 concentrations can do anything for ocean acidification. And while removing CO2 out of the atmosphere directly may well become a viable means to geoengineer the Earth’s climate and slow or reverse the increase in global mean surface temperature, the total amount of CO2 that needs to be removed is more than the entire amount of CO2 present in the atmosphere in any one year. And no CO2 removal technology will be effective unless humanity also stops emitting CO2 by burning fossil fuels.
Climate disruption is a hole, and before we consider how best to use geoengineering to build us a ladder out of the hole, we have to stop digging ourselves even deeper.
Atmospheric Chemistry and Physics
Environmental Research Letters