Introduction

Aerosol, as one of the most critical pollutants in the atmosphere, is harmful to environment, human health, and ecosystems1, and it is also considered as a short-lived climate forcer that has an impact on global and regional climate through interacting with radiation and clouds2. It is estimated that the global mean aerosol effective radiative forcing (ERF) at the top of the atmosphere (TOA) in 2019 relative to 1750 is −1.1 (−1.7 to −0.4) W m−2, with −0.22 (−0.47 to 0.04) W m−2 contributed from aerosol–radiation interactions and –0.84 (−1.45 to −0.25) W m−2 attributed to aerosol–cloud interactions. It leads to the global surface air temperature changed by −0.13 (−0.28 to −0.01) °C from aerosol-radiation interactions and −0.38 (−0.77 to −0.12) °C from aerosol-cloud interactions3.

Aerosols have been changing considerably throughout the globe. In the 1980s, clean air actions were implemented in North America and Europe and emissions of aerosols and their precursors have decreased since then4,5. As a consequence, near-surface aerosol concentrations and aerosol optical depth (AOD) over North America and Europe have declined during the past decades reported from long-term ground observations, satellite retrievals and model simulations6,7,8,9,10,11,12,13,14. Unlike developed countries in Europe and America, other developing regions such as South Asia continued to maintain high levels of emissions, with the aerosol levels still showing a significant increase5,15. In China, Air Pollution Prevention and Control Action Plan was promulgated in 2013 to deal with the serious air pollution issues. Subsequently, the emissions of aerosols and precursors have been reduced and aerosol concentrations have declined over China16,17,18. Ground observations indicated that the near-surface PM2.5 (particulate matter with diameter less than 2.5 μm) concentrations in China decreased by about 33.3% during 2013–201719.

Changes in aerosol, as an important climate forcer, come with local and regional climate impacts. During the period 1980–2010, aerosol emissions reductions in the U.S. caused changes in aerosol direct radiative forcing (DRF) by 0.8 W m−2 and indirect radiative forcing (IRF) by 1.0 W m−2 over eastern U.S., resulting in a 0.35 °C warming in the U.S.20,21. It has also been estimated that aerosol decreases over Europe lead to a DRF change of 1.26 W m−2 between the 1980s and 1990s22. Based on the chemical transport model GEOS-Chem, Dang and Liao23 found a regional mean change in DRF of 1.18 W m−2 in 2017 relative to 2012 resulting from the reduction in aerosols over eastern China. Zheng et al.16 reported that the decreased aerosol emissions in China during 2006–2017 exerted an anomalous ERF of 0.48 ± 0.11 W m–2 and a surface air warming of 0.12 ± 0.02 °C in East Asia. Gao et al.18 also estimated a warming of 0.09 ± 0.10 °C in eastern China contributed by decreases in domestic aerosols related to clean air policies during 2013–2017.

Aerosols affect climate not only over regions where aerosols are emitted, but also over remote areas. Cowan & Cai24 attributed the decreased Asian monsoon rainfall to strengthened northerly flows related to non-Asian aerosols. Wang et al.25 found an anomalous cooling in summer over East Asia partly due to eastward cold advections in the midlatitudes of the Northern Hemisphere caused by European aerosol forcing based on global aerosol–climate model simulations. Mahmood & Li26 demonstrated that South Asian black carbon (BC) aerosol could change the regional pattern of East Asian summer monsoon rainfall in China related to a propagating wave train along the Asian upper tropospheric jet induced by an increase in South Asian BC. Liu et al.27 also found that the growth in South Asian aerosols during 2013–2017 increased surface air temperature over central-eastern China and enhanced East Asian summer monsoon strength through changing large-scale atmospheric circulations. Wilcox et al.28 showed significant climate responses across the midlatitude to Asian anthropogenic aerosols related to the aerosol-generated Rossby waves. Fahrenbach & Bollasina29 investigated climate impacts of COVID-19-related Asian aerosol emission reductions and discovered a Hemispheric-wide response because of stationary Rossby wave trains from East Asia. In addition, Westervelt et al.30, Acosta Navarro et al.31 and Kasoar et al.32 all studied climate responses to aerosols from individual source regions in the Northern Hemisphere and reported a wide range of temperature responses across the hemisphere.

Climate responses to external forcers including anthropogenic aerosols can be decomposed into fast and slow climate responses33,34,35. Fast climate responses refer to changes in climate variables associated with the rapid atmospheric adjustments before significant changes in surface temperature, which occur on the timescale from days to months. Slow climate responses occur on a longer timescale of years, which refer to changes in climate variables associated with slow oceanic processes. Many previous studies have investigated fast and slow climate responses, especially precipitation response, to anthropogenic aerosols34,35,36,37,38. Kvalevåg et al.36 suggested that the fast response in precipitation is strongly correlated with radiative forcing associated with atmospheric absorption, while the slow response mostly depends on radiative forcing at TOA. Samset et al.34 also found that the response in precipitation over a number of land regions is more dominated by rapid adjustments than temperature-driven slow processes. Zhang et al.35 analyzed the fast and slow responses of precipitation to BC and sulfate perturbations and found a decrease in global-mean precipitation attributed to fast responses to BC perturbation and slow responses to sulfate perturbation. A few studies also discussed fast and slow responses in monsoon system to aerosols33,39. For example, Wang et al.39 found that East Asian summer monsoon was weakened due to a decreased land-sea temperature contrast in the fast response to sulfate forcing, and this effect was strengthened by the subsequent tropospheric thermodynamic and dynamic structure changes due to the slow response.

With the fast climate responses considered alone, Gao et al.18 suggested that domestic reductions in aerosols and precursor emissions related to local clean air actions induced less significant warming over eastern China, in which the slow oceanic process was not included. In this study, inspired by Gao et al.18, we aim to quantify climate responses in China to changes in both domestic and foreign anthropogenic emissions of aerosols and their precursors during 2013–2019 using the Community Earth System Model version 1 (CESM1) and decompose the climate responses to the fast and slow components to improve the understanding of the physics underlying the responses.

Results

Changes in anthropogenic aerosols from 2013 to 2019

Since 2013 when clean air actions were taken in China, anthropogenic emissions of aerosols and their precursors showed great reductions over China (Supplementary Fig. 1). Sulfur dioxide, BC, and organic carbon emissions decreased by −34.2, −0.8, and −0.58 Tg year−1, respectively, over China during 2013–2019. Although the emission reductions in Europe and North America experienced an unexpected slowdown in the past decade11,40, reductions in aerosols and precursor emissions can also be observed in these two regions.

Drastic declines in near-surface PM2.5 concentrations (sum of sulfate, BC, POM, SOA, dust×0.1, and sea-salt×0.25 following Turnock et al.41) are observed over eastern China in both model results and observations (Fig. 1). The model and observations also indicate a significant reduction in near-surface PM2.5 concentrations over eastern U.S. and Europe, but the reduction in Europe and eastern U.S. is not as substantial as in China. The simulated and satellite AOD also show decreases in China, eastern U.S., and Europe (Supplementary Fig. 2). Please see the detailed model evaluation in the Methods section.

Fig. 1: Comparisons of near-surface PM2.5 concentrations between observations and simulations.
figure 1

Spatial distributions of observed (circles) and simulated (shades) annual mean near-surface PM2.5 concentrations (sum of sulfate, BC, POM, SOA, dust×0.1, and sea-salt×0.25) during 2013–2015 (a–c), 2017–2019 (d–f), and their differences (g–i, 2017–2019 minus 2013–2015) over China (a, d and g, CN), North America (b, e and h, NA), and Europe (c, f and i, EU). Normalized mean bias (NMB) and correlation coefficient (R) between observation and simulation are shown at the upper-left corner of each panel. NMB = 100% × ∑ (Modeli − Observationi)/ ∑ Observationi, where Modeli and Observationi are the modeled and observed values at site i, respectively.

Full size image

Climate responses to anthropogenic aerosol changes

As a short-lived climate forcer, aerosols affect climate by altering the earth’s radiation budget through interacting with radiation and clouds. The implementation of clean air actions in China resulted in notable reductions in anthropogenic aerosols and precursor emissions, which led to a decrease in near-surface PM2.5 concentrations, as illustrated above. The reductions in aerosols further caused an anomalous aerosol ERF of 0.64 W m−2 at TOA, averaged over China in 2019 relative to 2013, with the largest ERF anomaly exceeding 2 W m−2 in southern China (Fig. 2a). Concurrently, the emission reductions policies implemented in Europe and North America induced an anomalous ERF of 0.48 and 0.45 W m−2, respectively, over these two local regions. Note that, in PM2.5, both absorbing aerosol (i.e., BC) and scattering aerosols (i.e., sulfate, POM and SOA) decreased over China, Europe and North America. The positive ERF anomaly at TOA indicates that the climate impacts of aerosol changes during 2013–2019 are dominated by the scattering aerosols.

Fig. 2: Changes in effective radiative forcing (ERF) due to changes in anthropogenic emissions of aerosols and precursors.
figure 2

Spatial distributions of changes in annual mean ERF (W m−2) due to changes in (a) global, (b) China’s domestic and (c) foreign anthropogenic emissions of aerosols and precursors in 2019 relative to 2013. Changes in ERF are calculated as the differences in net radiative fluxes at the top of the atmosphere. The shaded areas indicate results are statistically significant at the 90% confidence level. Regional averages of the changes over China (CN), North America (NA) and Europe (EU) are noted near these regions.

Full size image

The positive ERF anomalies induced by aerosol reductions led to an increase in surface air temperatures in China, North America and Europe during 2013–2019. Regional averages of surface air temperatures in China, North America and Europe increased by 0.20, 0.15 and 0.14 °C, respectively (Fig. 3). Due to the emission reductions mainly occurring in the Northern Hemisphere, temperature increases in the Northern Hemisphere are much more pronounced than those in the Southern Hemisphere. Moreover, warming is not only limited to areas of emission reductions. The aerosol reduction-induced warming spreads across the Northern Hemisphere, including land and ocean regions, which has been reported in previous studies29,30,32. In addition, a faster warming in the Arctic than other regions related to Arctic amplification is also observed, which has been attributed to a combination of positive feedback mechanisms in the Arctic climate system42,43,44,45,46,47.

Fig. 3: Changes in surface air temperature and precipitation rate due to changes in anthropogenic emissions of aerosols and precursors.
figure 3

Zonal-mean (a and c) and spatial distributions of changes (b and d) in annual mean surface air temperature (°C, a and b) and precipitation rate (mm day−1, c and d) due to changes in global anthropogenic emissions of aerosols and precursors from 2013 to 2019. The shaded areas indicate results are statistically significant at the 90% confidence level. Regional averages of the annual mean surface air temperature changes over China (CN), North America (NA) and Europe (EU) are noted in the top panels.

Full size image

The warming induced by aerosol emissions reductions in the Northern Hemisphere influenced the hemispheric temperature contrasts, leading to a northward shift of the intertropical convergence zone occurred towards the warmer hemisphere48,49, as also depicted in Fig. 3. However, the precipitation changes over most of the continental regions are statistically insignificant, likely related to the small perturbation of aerosols during 2013–2019, especially the solar-absorbing particles, compared to the internal variability.

Responses in climate of China to domestic and foreign aerosol changes

In this section, climate responses over China to changes in global, domestic and foreign anthropogenic emissions of aerosols and precursors, and contributions from fast and slow processes are quantified, as shown in Fig. 4. A regional warming of 0.2 °C over China (Fig. 4a) is equally attributed to domestic (Fig. 4b) and foreign (Fig. 4c) emission reductions, i.e., each contributing to 0.1 °C of the increase in temperature, but the spatial warming patterns are different between the domestic and foreign contributions. Due to the reductions in China’s domestic emissions, aerosol decrease-induced warming (Fig. 4b) is contributed by both rapid atmospheric adjustments (Fig. 4e) and slow oceanic processes (Fig. 4h). Over eastern China, the fast climate response (Fig. 4e) dominates the warming caused by domestic aerosol reductions, in accordance with the positive ERF anomaly over this region (Fig. 2b). Although temperature increases are also located in western China due to the slow climate response (Fig. 4h), they are statistically insignificant and may be related to the internal variability of the model.

Fig. 4: Changes in surface air temperature due to changes in anthropogenic emissions of aerosols and precursors.
figure 4

Spatial distributions of total (a–c), fast (d–f) and slow (g–i) responses in annual mean surface air temperature (°C) to changes in global (a, d and g), domestic (b, e and h) and foreign (c, f and i) anthropogenic emissions of aerosols and precursors during 2013–2019. The shaded areas indicate results are statistically significant at the 90% confidence level. Regional averages of the responses over China are noted at the upper-right corner of each panel.

Full size image

It is intriguing to note that foreign aerosol changes significantly enhance the warming over China during 2013–2019 (Fig. 4c). Over eastern China, slow climate responses control the local warming from foreign contributions, with the maximum warming approaching 0.3 °C (Fig. 4i). Besides, the increasing aerosols during 2013–2019 in South Asia and Southeast Asia enhanced the transboundary aerosol transport into China (Supplementary Fig. 3), which causes a cooling over southern China through rapid atmospheric adjustments from the foreign contribution (Fig. 4f). It can be confirmed by the atmosphere-only simulation with emissions in South Asia and Southeast Asia alone changing from 2013 to 2019 (Supplementary Fig. 4). However, the insignificant ERF anomaly contributed by changes in foreign emissions (0.01 W m−2, Fig. 2c) cannot explain the overall warming in eastern China from foreign contributions. With changes in anthropogenic emissions of aerosols and precursors in North America and Europe alone, a 0.1 °C regional warming is also presented in China (Supplementary Fig. 5), which equals to that caused by changes in all foreign emissions. Also, the contribution of warming from emission changes over North America and Europe is mainly through slow oceanic processes (Supplementary Fig. 5c). It suggests that the aerosol reductions in the North America and Europe are the primary contributor to the foreign emission-induced warming in China during 2013–2019. This finding is consistent with previous studies21,30,31,32 that a perturbation in aerosol loading in the North America or Europe can exert a significant impact on air temperature in China.

The climate in China is also linked to foreign aerosol changes through teleconnections. Many studies have reported that the North Atlantic Oscillation (NAO) had significant downstream influences on East Asian climate, primarily through quasi-stationary wave propagation of upper-tropospheric anomalies along the Asian jet, along with positive temperature anomalies in East Asia during positive NAO phase50,51,52. The changes in foreign aerosols, especially aerosol reductions in the U.S. and Europe, increase the sea surface temperature (SST) over the North Atlantic Ocean (Supplementary Fig. 6), which is also likely to induce an anomalous warming over East Asia through anomalous wave trains propagation as the NAO influence.

Figure 5 shows total, fast and slow responses of 200 hPa eddy stream function and wave activity flux to changes in foreign anthropogenic emissions of aerosols and precursors, which have been widely used to examine teleconnection mechanisms53,54,55. Two anomalous wave trains are observed in the total responses, associated with anomalous warming in the Eastern U.S., North Atlantic Ocean, and Europe. One anomalous wave train propagates eastward from the Eastern U.S. to Northern China (via Eastern U.S.–North Atlantic Ocean–the Middle East– the Tibet Plateau–Northern China), while the other one originates from North Atlantic Ocean and Northern Europe, propagating through Kazakhstan to Northern China. In this study, positive stream function anomalies, originating from North America, North Atlantic Ocean, and Europe, propagate when the dynamic ocean is enabled (Fig. 5a), exhibiting remarkable similarities with the positive NAO phase51. This implies a teleconnection that links the increase in temperature over China to the reduction of aerosols in North America and Europe. The anomalous wave trains in the fast responses show different patterns than those in the total response, but are less statistically significant (Fig. 5b), while the anomalous wave trains in the slow responses are similar to those observed in the total responses (Fig. 5c). The spatial patterns of 500 hPa temperature, 200 hPa geopotential height, and 200 hPa eddy geopotential height exhibit noticeable similarities as the pattern of 200 hPa eddy stream function, characterized by positive anomalies in Eastern U.S., North Atlantic Ocean, Northern Europe, the Middle East, and Northern China (Fig. 6). These suggest that significant warming in Eastern U.S., North Atlantic Ocean and Europe in response to local aerosol emission reductions have significant remote climate influences via teleconnections when dynamic ocean processes are included, leading to a temperature increase in eastern China.

Fig. 5: Changes in 200 hPa eddy stream function and wave activity flux due to changes in foreign anthropogenic emissions of aerosols and precursors.
figure 5

Spatial distributions of (a) total, (b) fast and (c) slow responses in 200 hPa eddy stream function (105 m2 s−1, shades) and wave activity flux (m2 s−2, vectors) to changes in foreign anthropogenic emissions of aerosols and precursors. The shaded areas indicate results are statistically significant at the 90% confidence level. The stream function and wave activity flux are calculated according to Takaya & Nakamura78. The non-zonal/eddy anomalies are obtained by removing the zonal averages in each latitude53. Anomalous wave train propagation is shown in black arrows in the top panel.

Full size image
Fig. 6: Changes in 500 hPa temperature, 200 hPa geopotential height and 200 hPa eddy geopotential height due to changes in foreign anthropogenic emissions of aerosols and precursors.
figure 6

Spatial distributions of total (a–c), fast (d–f) and slow (g–i) responses in 500 hPa temperature (K, a, d and g), 200 hPa geopotential height (m, b, e and h) and 200 hPa eddy geopotential height (m, c, f and i) to changes in foreign (non-China) anthropogenic emissions of aerosols and precursors. The shaded areas indicate results are statistically significant at the 90% confidence level.

Full size image

The teleconnection can also be partly supported based on by the maximum covariance analysis (MCA)53,56,57 between the surface temperature (TS) over North America, the North Atlantic Ocean, and Europe and the 200 hPa geopotential height (GPH) during 1970–2020 from EAR-5 reanalysis. Explaining 69.3% of the covariance, the leading MCA mode shows the anomalous high TS over North America, the North Atlantic Ocean, and Europe is accompanied by the anomalous high GPH across Eurasia (Supplementary Fig. 7a), which is in agreement with the CESM1 simulations (Fig. 6b). The associated time series for this mode exhibit a strong correlation, with the correlation coefficient of 0.87 (Supplementary Fig. 7b). Although this mode in observations is more likely related to the global warming caused by the increases in greenhouse gases, the simulated anomalous high GPH induced by changes in foreign emissions between 2013 and 2019 resembling the GPH pattern in observations might contribute to the warming in Northern China.

Discussion

In recent years, Europe and North America have continuously reduced their anthropogenic emissions of aerosols and precursors, and China has also significantly reduced its emissions since the implementation of clean air policies in 2013. Several studies16,18,23 have reported that clean air actions in China in recent years exerted a local warming, but few of them considered the effects of simultaneous aerosol changes in foreign regions outside China. Additionally, both fast and slow processes can lead to climate change through different mechanisms, but most previous studies did not separate the two processes. This study quantifies the climate responses in China to changes in domestic and foreign emissions of aerosols and precursors during 2013–2019 and distinguishes the fast and slow climate responses using CESM1.

Anthropogenic aerosol concentrations have significantly decreased in China, North America, and Europe during 2013–2019, which further led to increases in surface air temperature across the Northern Hemisphere. The global aerosol changes induce a regional warming by 0.2 °C over China during 2013–2019, which on average is equally contributed by China’s domestic and foreign emission changes. Over eastern China where warming is the most significant, the domestic aerosol-induced warming is primarily due to fast climate responses (i.e., rapid atmospheric adjustments). Foreign aerosol changes, especially aerosol reductions in North America and Europe, contributed to an average warming of 0.1 °C in China mainly through slow climate responses (i.e., slow oceanic processes). The foreign aerosol-induced warming is related to anomalous wave train propagation in the Northern Hemisphere associated with anomalous warming in the Eastern U.S., North Atlantic Ocean, and Europe. The comparable magnitude of the influence of domestic and foreign aerosol changes on surface air temperature in China during 2013–2019 highlights the critical role of foreign aerosols in regulating climate in China, which also emphasizes the importance of cooperation and governance between countries in global climate mitigation.

There are several limitations and uncertainties in our study. Firstly, CESM1 significantly underestimates PM2.5 concentrations in China and Europe. This underestimation may result from a few factors, including but not limited to uncertainties in new particle formation, strong aerosol wet removal, coarse model resolution, and uncertainties in anthropogenic emissions of aerosols and precursor gases58,59,60. Consequently, the low bias in estimated aerosol could lead to an underestimation of the simulated climate responses. However, the climate responses in China to domestic or foreign emission perturbation may not be linear and CESM1 is more sensitive to anthropogenic forcings than many other climate models3,61 related to its strong aerosol-cloud interactions62. Therefore, the more accurate climate responses to the aerosol changes need further simulations with a correct AOD and concentration perturbation in further studies with more models63. Secondly, nitrate and ammonium, two major components of aerosols, are not explicitly treated in this model version, of which changes must have impacted the climate59,64. They need to be considered in future studies. Thirdly, 150-year fully-coupled equilibrium simulations were conducted in our study, while equilibration with the deep ocean can take thousands of years. Longer simulations may lead to more convincing results. Finally, our results rely on simulations using one single model, and future studies are warranted to employ multi-model ensemble simulations to reduce a potential model dependency.

Methods

Observational data and meteorological reanalysis

Ground-based observational data of PM2.5 concentrations in China, U.S. and Europe are derived from China National Environmental Monitoring Centre (CNEMC), Interagency Monitoring of Protected Visual Environments (IMPROVE) and European Monitoring and Evaluation Programme (EMEP), respectively. These networks provide daily near-surface air pollutant concentrations. AOD data are also obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) Deep Blue retrieval65. These observational data are applied to evaluate the model performance.

ERA-566, the fifth generation of the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis, is a comprehensive dataset with spatial resolution of 0.25 degrees. Monthly surface temperature (TS) and geopotential height (GPH) data from 1970 to 2020 are acquired from ERA-5 for the maximum covariance analysis in our study.

Model description and emissions

In this study, we simulate aerosol climate effects using the global earth system model CESM167. CESM is a comprehensive climate model that simulates the earth’s system. Its components include the Community Atmosphere Model (CAM) for atmospheric processes, the Community Land Model (CLM) for terrestrial processes, the Parallel Ocean Program (POP) for global ocean dynamics, and the Community Ice Code (CICE) for polar sea ice processes. It can represent various aspects of the earth’s environment and interactions within the climate system. The atmospheric component of CESM1 in this study is CAM version 5 (CAM5), which in this study applies a horizontal resolution of 1.9° latitude × 2.5° longitude and 30-level vertical layers. The four-mode version of the Modal Aerosol Module (MAM4) within CAM5 predicts aerosol species including sulfate (SO42−, generated from its precursor sulfur dioxide (SO2)), BC, primary organic matter (POM), secondary organic aerosol (SOA, generated from a lumped SOA gas precursor (SOAG)), mineral dust, and sea salt in four lognormal modes (i.e., Aitken, accumulation, coarse, and primary carbon modes)68. Sulfate aerosol is mainly produced from its precursor SO2 through gas-phase oxidation and aqueous-phase oxidation in bulk cloud water. The treatment of SOA is to assume fixed mass yields for precursor volatile organic compound, then directly emit this mass as primary aerosol particles. MAM adds one additional step of complexity by simulating a single lumped semi-volatile organics gas-phase species, which is referred to as SOAG in the model. The CAM5 model includes aerosol-radiation interaction in shortwave and longwave bands as well as aerosol-cloud interactions for stratiform clouds69. The double-moment formulation of Morrison and Gettelman70 and Gettelman et al.71 is employed to handle the microphysical processes of this stratiform cloud. It enables the prediction of both number and mass mixing ratios of cloud droplets and ice crystals, while the number and mass mixing ratios of rain and snow are diagnosed in the model. Rapid Radiative Transfer Model for General circulation models (RRTMG72) is configured in the model for radiative transfer calculations. Precipitation is diagnosed by the Morrison‐Gettelman cloud microphysics scheme version 1 (MG170,71). A few modifications for improving the aerosol convective transport and wet deposition are added in the model following Wang et al.73.

The global anthropogenic emissions of aerosols and precursors are from the latest Community Emissions Data System (CEDS) v_2021_04_215. Different from previous CEDS v_2016_07_26, which has large biases in regional emissions74, anthropogenic aerosols and precursors emissions were updated with country-level emission inventories of China, Europe and North America. It has now considered the substantial emissions reductions in China due to the clean air actions in recent years. Biomass burning emissions of aerosols and precursors are from CMIP6 (the Coupled Model Intercomparison Project Phase 6)75. Biogenic emissions are obtained from the Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGAN v2.1)76.

Experimental design

This study aims to quantify the impacts of the changes in domestic and foreign anthropogenic aerosols on climate in China from 2013 to 2019 and investigate the mechanisms of fast and slow climate responses. A series of model experiments are performed using CESM1, as showed in Table 1. ALL2013 is a baseline experiment with a fully-coupled model configuration, in which anthropogenic aerosols and precursors emissions are fixed at year 2013 over the entire globe. In CHN2019, anthropogenic emissions of aerosols and precursors over China are fixed at year 2019 but fixed at year 2013 over all other regions. In ALL2019, anthropogenic emissions of aerosols and precursors are fixed at year 2019 over the entire globe. An additional simulation USEU2019 is also performed, in which anthropogenic emissions of aerosols and precursors over North America and Europe are fixed at year 2019 but fixed at year 2013 over all other regions. All these fully-coupled simulations are run for at least 150 years, with the last 90 years being analyzed. For the purpose of minimizing the model noise, three ensemble members are conducted for each fully-coupled simulation by introducing temperature perturbations to the initial atmospheric conditions.

Table 1 Experimental design.
Full size table

Another group of atmosphere-only experiments driven by prescribed monthly SST and sea ice concentrations derived from ALL2013 are performed. The experiments are run for 100 years, with the last 90 years selected for analysis. The emission setup in the atmosphere-only experiments is identical to that in the corresponding fully-coupled experiments. These atmosphere-only simulations are used to quantify climate responses that are attributable to the fast processes and to estimate changes in ERF imposed by changes in anthropogenic emissions of aerosols and precursors. Note that, to identify the impacts of changes in emissions from South Asia and Southeast Asia through fast processes, one additional atmosphere-only experiment SASEA2019_FAST is also conducted, in which anthropogenic emissions of aerosols and precursors over South Asia and Southeast are fixed at year 2019 but fixed at year 2013 over all other regions.

Total climate responses to emissions changes are illustrated by the differences between each pair of the fully-coupled simulations. Total climate responses to changes in anthropogenic emissions of aerosols and precursors from the entire globe, China, non-China regions, as well as North America and Europe, during 2013–2019 are identified as the differences between ALL2013 and ALL2019, ALL2013 and CHN2019, CHN2019 and ALL2019, and ALL2013 and NAEU2019, respectively. Fast climate responses are calculated as the differences between each pair of the atmosphere-only simulations, while differences between total response and fast response are attributed to slow response.

Model evaluation

The simulated near-surface PM2.5 concentrations in China, North America and Europe in years 2013 and 2019 and the changes during this time period are compared with observations in Fig. 1. Note that, to minimize the meteorological influences on the aerosol observations, the observational data are averaged during 2013–2015 and 2017–2019. The use of smoothing for model evaluation has been employed in previous studies11,12. The model simulations can capture the observed spatial distribution of annual average near-surface PM2.5 concentrations in China and the changes during 2013–2019, with statistically significant correlation coefficients within the range of 0.5–0.6. However, the model considerably underestimates the PM2.5 concentrations and their declines by about 50–80% in China. Such biases have been documented in many previous studies18,60, which can be partly attributed to uncertainties in new particle formation, uncertainties in anthropogenic emissions of aerosols and precursor gases, and coarse model resolution58,59. The comparison between the European PM2.5 concentrations from the model and observations shows statistically significant correlation coefficients of 0.2–0.5, but still having a severe underestimation by 30%–60% in the annual mean concentrations and the decreases during 2013–2019. The underestimation in Europe was also reported in previous studies77. Both the model and observations indicate a significant reduction in near-surface PM2.5 concentrations over eastern U.S., and the model exhibits a good performance in reproducing the magnitude of the PM2.5 concentrations and changes with bias below 15%.

The model can well reproduce the spatial pattern with substantial decreases in AOD over eastern China and eastern U.S., but still underestimates the magnitude of the decrease in China by 69%, which suggests that climate effects due to aerosol reductions in China are likely underestimated by the model. Note that the 3-year smoothing of the observations cannot fully remove meteorological influences and the comparison here is only for reference. For example, in Europe, both the annual mean PM2.5 concentration and AOD are underestimated by 50–60% and the change in PM2.5 concentration during 2013–2019 is also underestimated by 60%, but the change in AOD is overestimated. Therefore, we speculate that the influence of aerosol reductions in Europe is likely underestimated in the CESM1 experiments.