Reduction in European anthropogenic aerosols and the weather conditions conducive to PM2.5 pollution in North China: a potential global teleconnection pathway

Frequent and severe PM2.5 pollution over China seriously harms natural environment and human health. Changes in meteorological conditions in recent decades have been recognized to contribute to the long-term increase in PM2.5 pollution in North China (NC). However, the dominant climatic factors driving the interdecadal changes of the weather conditions conducive to PM2.5 pollution remain unclear. Here we identify a potential global teleconnection mechanism: the decadal reduction in European aerosol emissions since the 1980s may have partially contributed to the interdecadal increase in weather conditions conducive to PM2.5 pollution in NC, measured by an Emission-weighted Air Stagnation Index (ASIE) that increases at a rate of 6.2% decade−1 (relative to the 1981–1985 level). By regression analysis, we show that the decreased European aerosol loadings can warm the lower atmosphere and induce anomalous ascending motion in Europe, which potentially stimulates two anomalous Rossby wave trains in the upper troposphere travelling eastward across Eurasia. The teleconnection patterns project on NC by weakening the near-surface horizontal dispersion, which may be favorable to the increase in local ASIE and air pollution build-up. The suggested mechanism is further supported by the results from a set of large-ensemble simulations, showing that the European aerosol emission decline since the 1980s excites similar local heating and ascending motion and leads to increasing trends of 0.1–0.5 μg m−3 (38 year)−1 in surface sulfate concentrations over most of NC. This proposed ‘West-to-East Aerosol-to-Aerosol’ teleconnection mechanism helps resolve opposite views on the impact of global versus local aerosol forcing on PM2.5 pollution weather in NC. The policy implication is that the sustained decline in European aerosol emissions in coming decades, in conjunction with unabated global and regional warming, could further exacerbate air pollution in NC, thus imposing stronger pressure to reduce local emission sources quicker and deeper.


Introduction
With rapid economic growth powered by intense fossil fuel (FF) use, China has encountered frequent and widespread air pollution in recent decades (Cai et al 2017. Notably the air pollution is due to a dense accumulation of particles with an aerodynamic diameter smaller than 2.5 µm (PM 2.5 ) near the ground, and it does great damage to human health (Pope and Dockery 2006, Rich et al 2012, Cohen et al 2017. A significant increase in anthropogenic aerosol (AA) emissions is the primary cause of intensified PM 2.5 pollution . However, the role of meteorology (e.g. horizontal dispersion, and planetary boundary layer (PBL) height) in the formation, accumulation, and dissipation of PM 2.5 pollution cannot be ignored. Here, we broadly refer to these meteorological conditions as 'PM 2.5 pollution weather' . The effects of PM 2.5 pollution weather have been clearly demonstrated at daily , seasonal (Wu et al 2016), or interannual time scales (Yu et al 2019). At the multi-decadal time scale, the focus of this analysis, it has been shown that global warming due to greenhouse gases (GHGs) have significantly contributed to the long-term increase in China's PM 2.5 pollution weather . The diverse mechanisms involved can include the reduction in surface relative humidity (Ding and Liu 2014), decline in Arctic seaice (Zou et al 2017), and weakening of the East Asian winter monsoon (Wu et al 2016).
But, climate change is not just GHG warming. Historically, AA forcing has imposed a net cooling effect on the climate system (Myhre et al 2013), and has been shown to significantly affect tropospheric circulation and global to regional precipitation (Bollasina et al 2011). It is less clear how the past changes in AA have affected air pollution related meteorology (i.e. PM 2.5 pollution weather). This is an intriguing issue because in addition to being a major climate forcing factor, AA is also the air pollutant itself. The mechanism of 'aerosol (forcing) to aerosol (response)' , mainly operating at a regional scale in South and East Asia (i.e. well documented interactions between aerosols and monsoon), serves as a key component in the intimately connected chemistryclimate loop (e.g. Lau et al 2016). A deeper understanding of how PM 2.5 pollution weather has changed in the recent past is more than an academic exploration; it is also informative for future policymaking, as governments worldwide are working to reduce local AA emissions to fight air pollution problems. The mitigated emissions will unquestionably contribute to the improvement of local air quality directly. However, it remains unclear whether, indirectly, the resultant changes in PM 2.5 pollution weather at a given location, due to local or remote AA emissions reduction, could enhance or offset the benefit of local AA emissions reduction.
Motivated by this, a limited number of studies focused on PM 2.5 pollution weather response to AA emission changes, but had provided opposite views (Hong et al 2020, Feng et al 2020b, Wang et al 2021. A recent regional chemistry-climate modeling study demonstrated that future reductions in local AA emissions over eastern China would lead to PBL changes that facilitate vertical ventilation to improve ground air quality (Hong et al 2020). In stark contrast, using large-ensemble experiments from an Earth system model that explicitly include aerosol-cloud interactions, Feng et al (2020b) indicated that in response to the decrease in global AA emissions, the horizonal dispersion would decline, thus worsening the PM 2.5 pollution weather conditions over eastern China.
Does the difference of the two future-looking modeling studies imply that non-local AA forcing far from Asia, such as in Europe, could have a strong influence on PM 2.5 pollution weather over China? Previous modeling analysis demonstrated that the reduction in European aerosol loadings led to local and Arctic warming (Acosta Navarro et al 2016) and exerted major influence on regional meteorology across Eurasia , Lewinschal et al 2019, Wang et al 2020a, primarily through changing the mid-latitude jet stream and temperature advection in the upper troposphere.
Here we put this hypothesis to the test by first observationally examining the potential linkage between the European AA reduction and the PM 2.5 pollution weather in North China (NC) for the period 1981-2018, during which the decrease in sulfur dioxide (SO 2 ) emissions in Europe was more than two times larger than both the decrease in North America and the increase in Asia (Lamarque et al 2010). We also analyze the results from a set of large-ensemble global model simulations, with specific regional AA changes as the single forcing since the 1980s, to further verify the observation-based empirical evidence.
To our knowledge, no observation-based empirical studies have been conducted to explore the possible teleconnection between aerosol forcing and aerosol (PM)-related air pollution (a global 'aerosol-toaerosol' connection) operating at a multi-decadal time scale, although there have been ample modeling and observational documentations on such a local 'aerosol-to-aerosol' connection operating at a time scale of days to weeks (e.g. Ding et al 2016, Gao et al 2016. Our findings here can improve the attribution of long-term changes in China's PM 2.5 pollution and thus better inform policymaking to achieve future clean air goals.

Observations
The historical SO 2 emissions (up to 2014) from the Community Emissions Data System (CEDS; Hoesly et al 2018) were used as a proxy for the temporal trend of AA in Europe, while SO 2 emissions for 2015-2018 were from the Shared Socio-economic Pathway (SSP) 2-4.5 (Rao et al 2017, Gidden et al 2019. According to Myhre et al (2017), European region was defined as 10 • W-40 • E, 35 • -70 • N (figure 3(a), box). Note that there were different trends in AA emissions between Western Europe and Central/Eastern Europe. More related to the magnitude of radiative forcing of AA, monthly aerosol optical depth (AOD) from the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) was used. The gridded products of annual surface PM 2.5 mass concentrations for 1998-2018 derived by combining AOD retrievals from multiple satellites with the Goddard Earth Observing System-Chem chemical transport model (Hammer et al 2020) were used here to depict the patterns and trends of PM 2.5 pollution over China during the past two decades, especially over the hotspot of NC region (defined as 111.9 • -119.4 • E, 32.5 • -41 • N; box in figure 7(c)).
To examine the relationship between European aerosols (measured by SO 2 emissions and AOD) and PM 2.5 concentrations in NC, we conducted regression/correlation analysis on related meteorological variables across Europe and Asia (using the fifth generation European Centre for Medium-Range Weather Forecasts monthly reanalysis (ERA5) data; Hersbach et al 2020) and the PM 2.5 pollution weather (indicated by Emission-weighted Air Stagnation Index, ASI E , details in the next section). To remove the interference of year-to-year fluctuation and to focus on the multi-decadal trend, all data are smoothed with a five year filter prior to any regression/correlation analysis. Statistical significance is assessed using a two-tailed Student's t-test.

ASI E
The ASI E serves as a synthetic meteorological proxy for quantifying the weather conditions conducive to PM 2.5 pollution. ASI E is expressed as: The ASI M depends on three key local meteorological factors: precipitation, PBL height, and surface wind speed within the PBL (Feng et al 2018), denoted as: where U (z) is the wind speed at the geopotential height z (unit: m); z 0 and z PBL are the geopotential heights of the surface and top of the PBL, respectively; r is the daily mean accumulated precipitation rate (unit: mm d −1 ); and δ (r) is equal to 1 and 0 for r ⩾ 1 and r < 1, respectively, depicting the high removing efficiency due to wet deposition. Note that a larger ASI E values (with smaller surface windspeed, lower PBL height, and lighter precipitation) indicated more favorable weather conditions for the build-up of PM 2.5 pollution. E s (x, y) is the normalized PM 2.5 -related emission factor E s (x, y), which was expressed as: where E s (x, y) denotes the overall impacts of emissions in grid cell (x, y) using the sum of nearby emissions E (x, y) weighted by distance and climatological wind fields, and the 'min' and 'max' denote the minimum and maximum of E s (x, y) over NC. E s includes the emission quantities of SO 2 , nitrogen oxides (NO x ), black carbon, organic carbon, terpenes, and other particle sources of PM 2.5 in the form of: The ASI E combines the effects of meteorology and the spatial distribution of emission (time invariant using the present-day level at year 2012). Notably, the time variation in ASI E depends only on the meteorology according to equation (1); thus, it is a PM 2.5 pollution weather indictor, rather than direct correspondence of PM 2.5 level. It has been shown that there are good spatial and temporal relationships between ASI E and observed PM 2.5 concentrations over timescales longer than one day (Feng et al 2020a).
The daily gridded ASI E is calculated from the MERRA-2 reanalysis product and then the monthly and annual ASI E were averaged from the daily values for the period 1981-2018. In order to examine if annual ASI E is relevant to frequency of 'PM 2.5 pollution weather' episodes, we first calculated the climatological value of the largest 10% of daily ASI E (i.e. ASI E _10%) over all grid cells within the entire NC domain and then we used the calculated ASI E _10% of 8.8 as a single threshold to define 'PM 2.5 pollution weather days' when local daily ASI E value exceeds the threshold. We obtain the number of 'pollution days' at each grid in a year (shown as a percentage) and the regional average of percentage measures the overall frequency of PM 2.5 pollution weather days in NC.

Wave ray tracing (WRT) method
The WRT method, developed from wave kinematics theory, has been used to detect the trajectory of Rossby wave propagation under a specified basic flow field (Li et al 2015(Li et al , 2019. By calculating a cubic ray equation with a specific starting point and initial wavenumber in the x-direction (i.e. west to east), WRT can compute the energy dispersion pathways of the Rossby wave. The method has been widely adopted to analyze the regional climate change due to global teleconnections (e.g. Shaman and Tziperman 2005, Li et al 2015, Scaife et al 2017, Li and Ruan 2018.
This study uses WRT to illustrate the physical linkages between the circulation anomalies directly caused by AA emission reduction in Europe and the PM 2.5 pollution weather variations over China.
The key variables for WRT calculation are the ERA5 monthly wind fields at 200 hPa. Following Li et al (2019), the climatological wind field was smoothed with the spectral triangular truncation at wavenumber 10 to remove small-scale disturbances. This ensemble has ten members for 1980-2020.

Chemistry-climate modeling
We obtain the climate responses to the regional changes in AA over the West by subtracting the results of the Fix_WestFF1920 from the CESM1 LE, and obtain the responses to the changes in aerosols over the East by subtracting the results of the Fix_EastFF1920 from the CESM1 LE. All results are based on the ensemble average of each run in order to isolate the climate responses to regional AA forcing from model simulated internal variability. Despite the inclusion of AA forcing over North America, figure 1(c) shows that the surface warming caused by the Western AA forcing is mainly located over Europe during 1981-2018, consistent with the larger reduction of AA emissions in Europe (figure 1(a)) and the simulated larger decrease in AOD over Europe (figure 1(b)) compared to North America.

Anti-correlation between European aerosols and ASI E in NC
The NC is a densely populated and economically developed region, which has experienced the highest concentrations and largest increases of PM 2.5 in China during the past few decades (supplementary figure 1 (available online at stacks.iop.org/ERL/16/ 104054/mmedia)). Figure 2(a) shows the changes in ASI E and percentage of days with larger than ASI E _10% value averaged over NC for the period 1981-2018. There are fluctuations during certain years: for example, the magnitude of the ASI E reaches a peak during 2013-2016, when NC experienced frequent and severe PM 2.5 pollution events (Cai et al 2017 despite general emission reduction . Overall, the ASI E and local PM 2.5 are correlated (Feng et al 2020a). There is a general increase in the ASI E in NC, indicating that the weather conditions favorable to PM 2.5 pollution in NC have gone through an interdecadal deterioration. The ASI E in NC increases at a rate of 6.2% decade −1 (P < 0.001; relative to the 1981-1985 mean). The largest growth in the ASI E occurs over places within NC having high air pollutant emissions (Feng et al 2020a). The interdecadal trend of the ASI E is consistent with finding of Zhang et al (2018), which defined air quality-related meteorological indices using a different method. There is a high correlation (r = 0.99) between the changes in annual mean ASI E and percentage of days with larger than ASI E _10%.
During the same period , AA emissions in Europe have decreased significantly ( figure 2(b), black line), which has led to an interdecadal decline in the AOD over Europe ( figure 2(b), blue line). The larger AODs during 1991-1995 (not shown) are primarily a consequence of the eruption of Mount Pinatubo volcano. Beside those few years, AOD is trending down, in line with the regional SO 2 emission decline. Clearly, there is a strong and intriguing anti-correlation between the increase in the ASI E or the percentage of days with larger than ASI E _10% in NC and the decrease in European AA. In particular, there is a sharp decline in AOD or SO 2 emissions but a steady rise in the ASI E or percentage of days with larger than ASI E _10% before 2000. The correlation coefficients of ASI E (percentage of days with larger than ASI E _10%) are −0.85 (−0.86) and −0.76 (−0.76), with SO 2 emissions and AOD respectively (supplementary table 1), all of which are significant at the 99.9% level. The high correlation could be indicative that   the decrease in European aerosol emissions in the recent decades have contributed to the long-term increases in weather conditions in NC conducive to annual mean PM 2.5 pollutions (ASI E ) and severe pollution episodes (percentage of days with larger than ASI E _10%). To establish this causality, next we explore the dynamic mechanisms that possibly connect the European aerosol forcing and the PM 2.5 pollution weather in NC using a regression method.

Local impact of reduction in European aerosol emissions
Change in aerosol emissions lead to a rapid response in the aerosol burden because of the short residence time (Textor et al 2006). There is a downward trend in AOD during 1981-2018 over almost all of Europe ( figure 3(a)). The largest decline in AOD is located over central Europe, with a trend exceeding −0.05 decade −1 . The decreased AOD lead to a weakening of atmospheric extinction of sunlight due to the interaction of aerosol particles with radiation and clouds, thereby increasing the net flux of surface solar radiation (SSR) (Myhre et al 2013). Indeed, the trend of clear-sky and all-sky SSR fluxes over Europe are similar in magnitude and pattern between, when expressed as regression coefficient with respect to the SO 2 emissions over Europe for 1981-2018 (figures 3(b) and (c)). Note that the regression coefficients are flipped in sign so that the positive values here always correspond to an increase in time. The temporal trend of SSR (i.e. simple linear regression with respect to time as done to AOD in figure 3(a)) yields similar results (figures 2(a) and (b)). The largest positive regression coefficients occur over the regions with the largest decline in AOD, especially for the clear-sky SSR ( figure 3(b)). There is also a high correlation between SSR and AOD or SO 2 emissions in Europe (supplementary table 1). The local surface radiation response is in line with the results in many previous studies that the decreased atmospheric aerosol loadings, rather than cloud changes in response to global warming, have been largely responsible for the surface brightening in Europe since the 1980s (e.g. The positive surface forcing also warms up the lower atmosphere ( figure 4(a)) and as a result, significant updraft anomalies (45 • -55 • N, black box in figure 4(b)) in the free atmosphere emerge particularly above the warming center with compensating descending motion at both flanks of the warming center (yellow to red regions in figure 4(b)). We further show in figure 5 that such a strong local response in SSR, surface and tropospheric temperature, and atmospheric circulation are all reasonably reproduced in the model experiment with the Western AA emissions as the single forcing, despite a slightly westward maximum center. Furthermore, when the same regression analysis is applied to the historical all-forcing driven CESM LE simulations (supplementary figure 4), we obtain similar results compared to the response derived as the difference between CESM1 LE and Fix_WestFF1920 simulations. Given the robust responses of local radiation, temperature, and circulation to European aerosol forcing as identified in the observational regression analysis and in the model simulations with specifically imposed regional forcings, we ask the next obvious question: what is the consequence downstream? Figure 6(a) show the annual anomalies of 200 hPa eddy geopotential heights (Z200e) linked to the SO 2 emissions over Europe for the period 1981-2018. The Z200e is defined as the deviation of geopotential height from its zonal mean. The anomalous Z200e over two downstream regions can be statistically connected to the anomalies over Europe in the upper troposphere (black box in figure 6(a)). One of the anomalous wave trains propagates eastward (i.e. the central Europe-Russia-North Pacific pattern, abbreviated as Russia wave train), and the other propagates southeastward from Europe to lower latitudes (i.e. the central Europe-West Asia-northern China-Western Pacific pattern, abbreviated as northern China wave train) (figure 6(a)). To further illustrate the potential physical connection between the local circulation anomalies in Europe and these downstream anomalies, WRT is used to compute the Rossby wave propagation pathway originating from anomalous Z200e in central Europe to downstream regions (supplementary figure 5). The propagation of the Rossby waves initiated from central Europe regions as directly diagnosed from 200 hPa wind fields follows the same two paths as the regression results based on Z200e revealed in figure 6(a).

Teleconnection between European aerosol emissions and the ASI E in NC
The diabatic heating induced anomalous ascending motion over Europe then excited the two anomalous Rossby wave trains across Eurasia, and they can also be identified in the mid to upper troposphere ( figure 6(b)), thereby effectively propagating the energy eastward and inducing a downstream response of large-scale circulation. Many earlier studies have identified the importance of diabatic heating in stimulating large-scale wave trains in the extratropics (e.g. Sardeshmukh and Hoskins 1988). The Russia wave train may lead to significant negative anomalous eddy geopotential heights in the upper troposphere of Russia ( figure 6(a)). Consequently, the pressure levels drop in the upper troposphere of the region. This increases the poleward and equatorward pressure gradient forces to the south and north of the anomalous Z200e center, respectively . An anomalous cyclone forms over the anomalous negative Z200e through the geostrophic balance between the pressure gradient force and Coriolis force (figure 6(c)), which then leads to an anomalous sinking motion beneath it (50 • -55 • N) ( figure 6(d)). The downward airflow diverges in the lower atmosphere, and an anomalous anticyclone appears at 850 hPa near 50 • N ( figure 7(b)). As a result, anomalous southeasterly winds to the south of the anticyclone may weaken the cold air intrusion from the higher latitudes that serves as a major dispersion driver for air pollution in NC . Indeed, there is a widespread reduction in near-surface wind speeds in large portion of NC ( figure 7(c)).
The northern China wave train may result in significant positive anomalous eddy geopotential heights in the upper troposphere of northern China and Mongolia (figures 6(a) and (b)). Thus, there are increases in equatorward and poleward pressure gradient forces to the south and north of the anomalous positive Z200e center, respectively. This leads to an anomalous anticyclone in the upper troposphere of those regions and an anomalous ascending motion beneath it (32 • -45 • N) (black box in figure 6(d)). The enhanced upward motion generates positive anomalous eddy geopotential heights at 500 hPa over the northern China-Mongolia region (figure 6(c)), thereby weakening the East Asian trough. Correspondingly, an anomalous cyclone forms at 850 hPa in eastern China between 32 • N and 45 • N (figure 7(b)) due to the enhanced ascending motion. This can lead to anomalous southerly winds in the lower atmosphere in the coastal areas of NC, and further intensifies the southeasterly wind anomalies (figure 7(b)) and weakens the near-surface wind speeds in NC (figure 7(c)) that are prevailingly northwesterly (figure 7(a)).
To test the dynamic mechanisms inferred from observational regression analysis, we further examine the simulated responses to aerosol forcing in Europe and North America derived as the difference between a pair of model experiments (see section 2). The model results essentially support the mechanisms suggested by regression analysis (figures 6(e) and (f)). Consistently, there are also two anomalous Rossby wave trains across Eurasia in the upper troposphere (figure 6(e)) due to the only perturbation of the aerosol forcing over the West, rather than global warming or internal variability. Compared to the regression results, the Russia pattern is reasonably reproduced, while the northern China pattern lies more southeastward. As a result, the simulated anomalous ascending motion (figure 6(f)) is further at the south of NC compared to the observation. A more concrete modeling evidence is the simulated increasing trends of 0.1-0.5 µg m −3 (38 year) −1 in surface sulfate concentration and of 0.001-0.006 (38 year) −1 in AOD in most of NC (figures 8(a) and (b)), despite invariant local aerosol emissions. However, the increases in sulfate and AOD levels in NC in response to the decrease in European aerosols are an order of magnitude smaller than those in response to the increase in Asian aerosols (figures 8(c), (d) and supplementary figure 6). This, unsurprisingly, indicates that the significant increase in AA emissions is the main cause of long-term increase in PM 2.5 pollution in China . Comparing the changes driven by meteorology and driven by local emissions (figure 8 top row vs bottom row) help contextualize the influence of the proposed teleconnection mechanism here, which plays a relatively small but non-negligible role.
The teleconnection patterns have a potential for weakening the horizontal dispersion of air pollutants in NC. Indeed, the long-term increase in the ASI E in NC is found to be primarily due to the decreases in near-surface wind speeds (i.e. horizontal ventilation) and decrease in PBL height (in MERRA-2;Feng et al 2020a). Other studies also reported that PM 2.5 pollution over NC is often accompanied by anomalous southerly winds in the lower atmosphere, decreased wind speeds, and a weakened East Asian trough (Chen and Wang 2015). The teleconnection can also suppress the downward transport of westerly momentum in NC, which not only preserves the inversion boundary layer but also blocks the dry and clean air from the upper levels (Yin et al 2021).

Conclusions and discussions
We examine a potential linkage of decreased European AA emissions and the weather conditions conducive to PM 2.5 pollution in NC for the period 1981-2018. Our results show that the decreased European AA emissions since the 1980s may have partially contributed to the interdecadal increase in both weather conditions in NC conducive to annual average PM 2.5 pollution (ASI E ) and severe pollution episodes (percentage of days with larger than ASI E _10%), primarily by modulating atmospheric circulation patterns across Eurasia (figure 9). Our regression analysis suggests that the decreased European AA emissions may explain the increase in SSR and the lower atmosphere warming over Europe. Two anomalous Rossby wave trains in the upper troposphere are stimulated across Eurasia due to the enhanced local diabatic heating and anomalous ascending motion in Europe. The teleconnection patterns then weaken the near-surface horizontal dispersion in NC, which may be favorable to the increase in local ASI E and air pollution build-up.
The suggested mechanism is further supported by the results from a pair of large-ensemble simulations with AA changes over the West as the single forcing, which is shown to excite similar local heating and ascending motion over Europe and West-to-East teleconnection patterns (despite a slightly different maximum center), and to increase the sulfate levels in NC. The regression analysis on CESM1 LE simulations driven by all historical forcing provides more supporting evidence that the regression analysis applied to reanalysis, while not conclusive, is indicative of some causal relationship. However, it should be emphasized that the significant increase in local AA emissions is the primary cause of long-term increase in PM 2.5 pollution in China.
Overall, the proposed causality is motivated from the strong correlation as indicated in figure 2 and supplementary table 1, and it is further strengthened by multiple lines of evidence: (a) statistical analyses that suggest the physical mechanisms connecting local heating in Europe to the circulation anomaly over NC favorable to local pollution build up; (b) the model sensitivity experiments in which the European aerosol reduction is introduced as the perturbation and yields a similar response, not only in the simulated dynamical fields but also in the simulated PM and AOD levels over NC; and (c) the same type of regression analysis applied to the historical model runs (supplementary figure 4), which produces qualitatively similar linkage demonstrated by the control-perturbation analysis in (b).
The implication of this study is multifold. This proposed 'West-to-East Aerosol-to-Aerosol' teleconnection mechanism helps resolve opposite views on the impact of global and local air pollution on PM 2.5 pollution weather in NC (i.e. Hong et al 2020, Feng et al 2020b. Europe is continuing to improve air quality, which will undoubtedly lead to a sustained decline of AA emissions in coming decades (Gidden et al 2019). Our results also imply that the remote climate response to future decreases in European AA emissions could partially offset the benefits of air quality improvement measures in NC.
Because of the distinct climate responses to AA in different models , we acknowledge that similar dedicated model experiments from other chemistry-climate models should be conducted and closely examined to further test the hypotheses here. Ideally, some dedicated runs should be performed with single forcing of 1980-2020 aerosol changes (in the West and in the East) to avoid the practice of subtraction in deriving climate response as done here. This additional test is useful because in principle, the different background climate conditions could affect how the local PM concentrations over China respond to remote aerosol forcing over Europe.
Note that the reduction in European AA affects the PM 2.5 pollution in China through both the decreased aerosol transport (although weak as indicated in figure 6(c)) and the increased ASI E . To fully separate the two factors, there is a need to conduct dedicated model experiments with meteorological and chemical fields separately prescribed to a regional chemistry-climate model over China. These too can be important research efforts using multiple models.