Climate change threatens to redistribute and endanger ecosystems worldwide, and these threats persist even under scenarios of ambitious policies to reduce greenhouse gas emissions. These and other climate change impacts motivate the study of climate interventions: methods to directly intervene in the Earth system to reduce climate risk alongside traditional mitigation and adaptation methods. It is important to emphasize that interventions are not a substitute for climate mitigation! However, they could have complementary goals by reducing impacts while mitigation efforts are ongoing.

One specific hypothetical intervention is stratospheric aerosol injection (SAI), where reflective particles would be placed into the upper atmosphere to reflect a small portion of sunlight away from Earth’s surface to limit warming or cool the planet. SAI is viable with near-future technology, and could act quickly – in an aggressive intervention scenario, it could be used to reduce temperatures within a few years after deployment.

Some researchers study SAI with climate model simulations to assess impacts without affecting the real world. One new set of simulations (“Assessing Responses and Impacts of Solar climate intervention on the Earth system with stratospheric aerosol injection,” or “ARISE” for short) produced by the National Center for Atmospheric Research includes a scenario that maintained global temperature with SAI and scenarios that cooled global temperatures rapidly after deployment. This led our team to wonder: what would the implications of rapid cooling be for global ecosystems?

One way to display the rate of temperature change relevant to ecosystems is through the climate speed: the magnitude of the ratio of the rate of temperature change to the spatial gradient of temperature. In other words, climate speed expresses how fast temperature conditions move with time, and faster climate speeds are harder for species to keep up with or adapt to. Researchers have applied climate speeds to SAI modeling before, with one key finding being a risk of rapid climate speeds from warming if SAI were stopped while greenhouse gas concentrations are high (“termination shock”). 

The ARISE simulations allow us to build on this work by connecting ecological risk to strategic logic: the specific choices that define a scenario. In ARISE-1.5, SAI is deployed in the year 2035 to maintain temperature at 1.5 ˚C above pre-industrial conditions. A second scenario, ARISE-DelayedStart, has a similar global target but deployment begins 10 years later in 2045. These scenarios allow us to tie the outcomes in the model simulations directly to this delay. We compare these results to baselines of no-SAI climate change with moderate mitigation (Shared Socioeconomic Pathway 2-4.5 [SSP2-4.5], consistent with present policy), and the pre-industrial evolution of the climate system (a “Last Millennium” simulation portraying years 850-1849).

The differences between the scenarios were striking. SSP2-4.5 (Figure 2a, 2b) showed large climate speeds, which emerged from two mechanisms. First, rapid changes in temperature cause large climate speeds–as in the Arctic, which is warming faster than the rest of the globe. Alternatively, when the spatial gradient of temperature is small even a small change in temperature implies species would have to move a large distance to keep up, which is particularly evident in tropical oceans.

In contrast, climate speeds were small in the Last Millennium simulation (Figure 2c, 2d) reflecting the relatively small influence and slow evolution of natural forcings and climate variability over the pre-industrial period. Climate speeds are similarly small in the ARISE-1.5 scenario (Figure 2e, 2f), particularly over land. Thus, SAI scenarios that maintain temperature have the potential to reduce ecological risk from climate speeds due to climate warming.

However, in the ARISE-DelayedStart simulation (Figure 2g, 2h), where deployment occurs similarly to ARISE-1.5 but ten years later in 2045, climate speeds were the largest yet. The rapid cooling necessary to compensate for the additional warming during the delay resulted in climate speeds as large or larger than the no-SAI SSP2-4.5 simulation. We termed this phenomenon “deployment shock.”

Each ARISE scenario contains ten different ensemble member simulations that depict equally-plausible representations of climate variability. Analyzing these members helps further illustrate the scale of the global response to the ARISE-DelayedStart intervention (Figure 3). On the planetary scale, ARISE-1.5 and the Last Millennium are statistically indistinguishable from each other; however, there is no overlap between these simulations and ARISE-DelayedStart! Even the ensemble member with the smallest speeds in ARISE-DelayedStart is beyond the member with the maximum speeds in either ARISE-1.5 or the Last Millennium. 

In the article, we find a consistent picture when considering a broader range of more scenarios in context to their ecological risk relative to simulated pre-industrial conditions. In climate change scenarios, global ecological risk increases directly with greater warming; however, even the most ambitious mitigation scenario still increases risk beyond pre-industrial conditions. For SAI scenarios with rapid temperature reduction, the deployment shock always increases risk relative to SSP2-4.5. Each scenario had unique strategic logic: deployment shock could emerge from the aforementioned 10-year delay, the intentional choice of a lower temperature target, or the earlier breaching of a target due to high climate sensitivity. In contrast, scenarios where temperature is maintained with SAI were the only cases where risk was within pre-industrial conditions. But, superficially small changes to these scenarios – such as a 10-year delay – can greatly increase ecological risk.

Our results help inform future climate intervention scenario design. In order to avoid deployment shock, temperature targets that require global cooling must be achieved gradually. For instance, a lower temperature target could be obtained if an intervention were slowly increased over decades to ensure climate speeds were never large. Research is increasingly identifying risks such as deployment shock tied to specific features of strategies. We believe future climate intervention scenarios should transparently document these risks and their prioritization in the scenario design process.