Explainer: How human-caused aerosols are ‘masking’ global warming

Global SO2 emissions under different SSP baseline and mitigation pathways compared to observed SO2 emissions from CEDS. Credit: Glen Peters.

Given differences across modeling groups, it is hard to infer too much about which SSP scenario is most in line with real-world SO2 emissions. However, it is worth noting that the current SSPs do not include a scenario where SO2 emissions continue to rapidly decline while emissions of CO2 and other GHGs increase.

Interestingly, the best-estimate cooling effect from sulphur dioxide is more or less counterbalanced by the warming effect of methane emissions today. As a result, scenarios where all GHG emissions are brought to zero do not result in sustained additional warming due to unmasking from declining aerosols. 

However, if CO2 emissions alone were reduced to zero, while non-CO2 emissions were held constant, cutting global aerosol emissions to zero would result in between 0.2°C and 1.2°C of additional warming.

This means that aerosol emissions represent something of a wildcard for future warming over the 21st century. Continued rapid reductions in SO2 emissions will contribute to an acceleration in the rate of global warming in the coming years.

Carbon Brief used the FaIR climate model to determine the effects of aerosol emissions on the climate, building on the work of Dr Chris Smith. Runs were done using the constrained ensemble approach using “fair-calibrate v1.4.”1 to be consistent with the IPCC AR6 parameter range. More details on the constrained ensemble approach can be found in Smith et al. (2024).

Figures showing the global mean surface temperature impact of different climate forcings in isolation were performed by calculating the difference between all-forcing runs and runs where a single forcing (e.g. from GHG emissions) was removed, following the approach used to generate Figure 7.8 in the IPCC AR6 climate science report.

IMO 2020 forcing estimates were taken from the following studies published in the peer-reviewed literature over the past two years:

IMO 2020 global average surface temperature changes were calculated by running 841 different FaIR simulations for each of the different forcing estimates identified in the literature, which is the default setting for the FaiR constrained ensemble to provide a range of results consistent with the IPCC AR6 parameter range.

This produced 6,728 total simulations, from which a central (50th percentile) estimate and uncertainty range (5th to 95th percentile) were calculated. 

These results were further validated by comparing them to the Earth system model-based estimates in individual studies where near-term global average surface temperature change estimates were provided (Yoshika et al. (2024); Quaglia and Visioni (2024); Gettelman et al. (2024); Jordan and Henry (2024); Watson-Parris et al. (2024); and Hansen et al. (2025). 

The results of each of these studies were within the range of FaIR based estimates for the respective study’s radiative forcing – and generally quite close to FaIR’s median estimate for that study, as shown in the table below. 

Comparison between Carbon Brief’s FaIR-based warming estimates of 2025 warming for a given study’s specified radiative forcing and the published near-term warming estimates from those studies. Note that studies lacking a global average surface temperature change estimate are excluded from this table, and that the published estimates may reflect years beyond 2025 when additional future warming is expected as the climate equilibrates to the change in forcing. The uncertainty range in published estimates may reflect uncertainty in the forcing as a result of emissions in addition to the climate response to the forcing changes.

It is worth noting that the uncertainties associated with converting SO2 forcing estimates to warming outcomes are generally much smaller than converting SO2 emissions into warming outcomes.

The effect of Chinese SO2 reductions were based on a comparison of two scenarios. The first is where Chinese SO2 emissions remained constant at their peak (2007) levels and did not decline. The second is where Chinese emissions followed observational estimates from CEDS between 2005 and 2022 and then remained constant at 2022 levels thereafter (which represents a conservative assumption that likely underestimates future effects of SO2 emissions declines on global temperatures given the strong downward trend). Global average surface temperature changes were calculated by running 841 different FaIR simulations in emissions mode for two scenarios and analysing the difference between the two.

The resulting estimate of 0.06°C (0.02°C to 0.13°C) warming by 2025 was validated by comparing it to the Samset et al. (2025) preprint, which finds a nearly identical best estimate of 0.07°C (0.02°C to 0.12°C) using a large set of simulations from eight different Earth system models.

The effects of the rest of the world’s SO2 declines were estimated using the same approach used for Chinese SO2 emissions, using CEDS emissions data. International shipping and aviation aerosols were excluded from the rest of the world estimate as to not double count IMO 2020 effects.

This story was published with permission from Carbon Brief.