Regional and global temperature response to anthropogenic SO<sub>2</sub> emissions from China in three climate models

Abstract

Abstract. We use the HadGEM3-GA4, CESM1, and GISS ModelE2 climate models to investigate the global and regional aerosol burden, radiative flux, and surface temperature responses to removing anthropogenic sulfur dioxide (SO2) emissions from China. We find that the models differ by up to a factor of 6 in the simulated change in aerosol optical depth (AOD) and shortwave radiative flux over China that results from reduced sulfate aerosol, leading to a large range of magnitudes in the regional and global temperature responses. Two of the three models simulate a near-ubiquitous hemispheric warming due to the regional SO2 removal, with similarities in the local and remote pattern of response, but overall with a substantially different magnitude. The third model simulates almost no significant temperature response. We attribute the discrepancies in the response to a combination of substantial differences in the chemical conversion of SO2 to sulfate, translation of sulfate mass into AOD, cloud radiative interactions, and differences in the radiative forcing efficiency of sulfate aerosol in the models. The model with the strongest response (HadGEM3-GA4) compares best with observations of AOD regionally, however the other two models compare similarly (albeit poorly) and still disagree substantially in their simulated climate response, indicating that total AOD observations are far from sufficient to determine which model response is more plausible. Our results highlight that there remains a large uncertainty in the representation of both aerosol chemistry as well as direct and indirect aerosol radiative effects in current climate models, and reinforces that caution must be applied when interpreting the results of modelling studies of aerosol influences on climate. Model studies that implicate aerosols in climate responses should ideally explore a range of radiative forcing strengths representative of this uncertainty, in addition to thoroughly evaluating the models used against observations.

Multimodel Surface Temperature Responses to Removal of U.S. Sulfur Dioxide Emissions

“…AOD changes are difficult to interpret as they show some of the least robust responses and least robust sensitivities. Kasoar et al (2016) also see a factor of 6 difference in AOD responses between the models they examined which included versions of the GISS and CESM models used here; furthermore, in that study the CESM AOD response was stronger than GISS in the source change region (China). We find here a stronger AOD response in GISS than in the CESM models in the source change region (United States).…”

“…Our study is a natural extension of those studies, intended to identify the full geographical temperature response resulting from regional (instead of latitudinal bands) emission reductions. Kasoar et al (2016) perform a similar study for SO 2 emissions from China. They find that aerosol optical depth (AOD) response to SO 2 emissions in China in the Hadley Centre Global.…”

Global source attribution of sulfate concentration and direct and indirect radiative forcing

“…The bias can also result from the fact that the site measurements are point observations, while the model results are grid-cell average that does not consider subgrid aerosol variations (Qian et al, 2010;. In addition, different models show large discrepancies in simulating sulfate over China (Kasoar et al, 2016). The underestimation of sulfate in China can lead to an underestimation of source contribution from East Asia of sulfate concentrations, direct and indirect radiative forcing of sulfate and forcing efficiencies of sulfate.…”

Multimodel precipitation responses to removal of U.S. sulfur dioxide emissions

“…Although there is some initial evidence that regional aerosol forcing can affect local and remote precipitation [Leibensperger et al, 2012;Shindell et al, 2012;Westervelt et al, 2015;Kasoar et al, 2016], additional work is needed particularly to establish statistical significance, determine robustness across multiple models, and understand physical mechanisms. This work aims to address this knowledge gap by estimating the impact of U.S.…”