Aerosols are the main air pollutants that contribute to excess morbidity and premature mortality by damaging cardiovascular and respiratory systems (Lelieveld et al., 2019). They also have adverse effects on atmospheric visibility for road and air traffic (Vautard et al., 2009). During the 1952 London Great Smog, air pollution associated with extremely high level of aerosols caused thousands of premature deaths (Bell and Davis, 2001), which resulted in a number of air quality legislative changes for reducing air pollution in Europe (Brimblecombe, 2006).

Besides the health and environment effects, aerosols can significantly impact regional and global climate by perturbing the Earth's radiation fluxes and influencing cloud microphysics (Boucher et al., 2013). Globally, anthropogenic aerosols exert a net cooling effect in the Earth system and have dampened the warming driven by greenhouse gases since the preindustrial era. Due to a strong surface albedo feedback over polar regions, the per unit aerosol emission from western Europe was reported to have a greater cooling effect than any other major source regions of the globe (Persad and Caldeira, 2018), revealing the importance of understanding aerosol variations in Europe.

Significant reductions in near-surface aerosol concentrations and aerosol optical depth (AOD) have been observed in Europe during the last few decades from long-term station measurements and satellite retrievals (de Meij et al., 2012; Tørseth et al., 2012). The decrease in aerosols has been considered as a cause of the increase in surface solar radiation over Europe since the 1980s (Wild, 2009), as well as a contributor to the eastern European warming (Vautard et al., 2009), Arctic amplification (Acosta Navarro et al., 2016), and increased atmospheric visibility over Europe (Stjern et al., 2011) during the past 3 decades.

The decrease in aerosols over Europe was mainly attributed to continuous reductions in local European anthropogenic emissions of aerosols and precursor gases since the 1980s (Smith et al., 2011), which are a result of legislation for improving air quality. In addition to local emissions, aerosol levels can also be affected by aerosol transport at continental scales (Zhang et al., 2017; Yang et al., 2018a). Aerosol emissions in major economic regions of the world have been changing rapidly during the past few decades owing to economic development and environmental measures. North America has started reducing emissions since the 1980s, and emissions in Russia also showed decreasing trends after the dissolution of the Soviet Union (Smith et al., 2011). In the meantime, aerosol emissions from East Asia and South Asia have largely increased due to economic growth, although emissions in China have been undergoing a remarkable reduction in recent years as a result of strict air quality regulations (Streets et al., 2000; Li et al., 2017). It is important to understand the relative roles of local emissions and regional transport in affecting long-term variations in aerosols in Europe from both air quality and climate perspectives.

Source apportionment is useful for quantifying contributions to aerosols from specific source regions and/or sectors and is beneficial to the emission control strategies. The traditional method of examining the source–receptor relationship in aerosol models is to zero out or perturb a certain percent of emissions from a given source region or sector in parallel sensitivity simulations along with a baseline simulation, and it has been used in many studies to examine source contributions of particulate matter (PM) in Europe from different sectors (e.g., Sartelet et al., 2012; Tagaris et al., 2015; Aksoyoglu et al., 2016). Recently, source region contributions to European CO and O3 levels, as well as global and regional aerosol radiative forcing, were examined under the Hemispheric Transport of Air Pollution model experiment phase 2 (HTAP2) protocol, in which sensitivity simulations were conducted with reductions in anthropogenic emissions from different source regions by 20 % (Stjern et al., 2016; Jonson et al., 2018). This method suffers from a large computational cost for the excessive model simulations when estimating contributions from a large number of sources, and contributions from all sources do not sum up to 100 % of the total concentration in the default simulation (Koo et al., 2009; Wang et al., 2014).

The explicit aerosol tagging method, which simultaneously tracks contributions from many different sources, is a useful tool for assessing the source–receptor relationship of aerosols. This method has previously been adopted in regional air quality models such as CAMx (the Comprehensive Air quality Model with Extensions) and CMAQ (the Community Multiscale Air Quality model). Using regional air quality models with aerosol tagging, contributions from different source sectors and local/regional sources to European PM and its health impact were studied (Brandt et al., 2013; Skyllakou et al., 2014; Karamchandani et al., 2017). However, due to the limitation in domain size of regional air quality models, contributions of intercontinental transport from sources outside the domain are difficult to account for.

Anthropogenic emissions of aerosols and their precursor gases from different economic regions of the world have changed substantially during the past few decades. Very few studies have examined the source apportionment of aerosols in Europe coming from sources all over the changing world. In this study, source attributions of concentrations, column burden, optical depth of aerosols in four major areas of Europe from 16 source regions of the globe over 1980–2018 are quantified. This is facilitated by the explicit aerosol source tagging technique that was recently implemented in a global aerosol–climate model (CAM5-EAST; Community Atmosphere Model version 5 with Explicit Aerosol Source Tagging). This technique has lately been used to examine the source attribution of aerosol trends in China and the United States during 1980–2014 (Yang et al., 2018a, b). The source apportionment analysis is extended to the year 2018 using the Shared Socioeconomic Pathways (SSPs) scenario, with a focus on Europe here.

The CAM5-EAST model, along with the aerosol source tagging technique, and aerosol emissions are described in Sect. 2. Section 3 evaluates the model performance in simulating aerosols in Europe. Section 4 shows the analysis of source–receptor relationships of aerosols in Europe on a climatological average. Source contributions to long-term variations in European aerosols and their direct radiative forcing (DRF) during 1980–2018, as well as future forcing prediction, are provided in Sect. 5. Section 6 summarizes these results and conclusions.

EMEP (European Monitoring and Evaluation Programme, http://www.emep.int, last access: 27 February 2020) networks provide daily near-surface aerosol concentrations in Europe. The annual mean of daily observations is used to evaluate the model performance in this study. Compared to the observational data from EMEP networks during 2010–2014, CAM5-EAST can reproduce well the spatial distribution and magnitude of aerosol components with normalized mean biases (NMBs) of -14 % to -23 % and correlation coefficients (R) in a range of 0.43–0.62 for sulfate, BC, and organic carbon (OC; derived from POA and SOA from the model results) (Fig. 2).

Figure 3 shows the time series of annual mean near-surface sulfate, BC, and OC concentrations averaged over EMEP sites in Europe and the corresponding model values during 1993–2018. Variations in near-surface sulfate concentrations are consistent between the model and observations, with R values higher than 0.9. The observed variations in BC and OC concentrations in Europe are represented in the simulation, with R values of 0.52 and 0.65, respectively. However, the observed high values of BC and OC concentrations are not captured by the model, probably because very few data were available before 2010, and, therefore, any difference between model and observation cannot be smoothed out through the spatial average. This is also indicated by the large spatial variation in BC and OC concentrations before 2010. Nevertheless, the modeled concentrations are still within the range of observations. Note that the number of sites used for the spatial average in Fig. 3 is different from year to year, and thus the variation or trend here does not represent that over a subregion or the entire Europe.

The modeled AOD is evaluated against the AERONET (Aerosol Robotic Network; https://aeronet.gsfc.nasa.gov, last access: 27 February 2020) data in Fig. 8. Both the modeled and observed AOD show decreasing trends during 2001–2018. The model underestimates AOD in all four subregions of Europe, probably due to the lack of nitrate aerosol. The variations in AOD in western Europe (combined NW and SW Europe) are well predicted, with R values of about 0.75, but the model barely reproduces the AOD variations in E Europe and the GTC region (R<0.5). The difference in the interannual variation in AOD between the model simulation and observation can be caused by many factors such as aerosol emissions, aerosol parameterizations in the model, the aerosol mixing state, inaccurate meteorological fields from reanalysis data, and biases in measurements. However, identifying the contribution of each factor to the difference is beyond the scope of this paper.

Based on the tagging technique in CAM5-EAST, near-surface concentrations of total sulfate-BC-POA can be attributed to emissions within and outside Europe, as shown in Fig. 4a and b, and the relative contributions in percentage are given in Fig. 4c and d. Averaged over 2010–2018, due to the relatively high local emissions, annual mean sulfate-BC-POA concentrations contributed to by European emissions show peak values of 4 µg m-3 in E Europe. The slight increase in SO2 emission from the GTC region (Fig. S1), which is opposite to the decreases in the other three subregions of Europe, also leads to high concentrations in GTC, with 2–4 µg m-3 contributed by European emissions. Due to the atmospheric transport from surrounding regions including northern Africa, the Middle East, and Russia–Belarus–Ukraine, non-European emissions account for 0.5–1 µg m-3 over SW Europe, E Europe, and the GTC area. Overall, local European emissions are the dominant sources of sulfate-BC-POA near-surface concentrations in Europe, with contributions larger than 80 % (60 %) in central areas (most of Europe). Non-European emissions are responsible for 30 %–50 % of the near-surface concentrations near the coastal areas and boundaries of the Europe that are easily influenced by aerosol regional transport.

Figure 5 illustrates the source contributions in percentage of emissions from different regions of the globe to near-surface aerosol concentrations and column burdens over the four subregions of Europe averaged over 2010–2018. Different aerosols have fairly different local/remote source attributions depending on the local to remote emission ratio and transport efficiency. European emissions explain 54 %–68 % of near-surface sulfate concentrations over the four subregions of Europe, with the largest local contribution in E Europe due to the relatively high emission rate. The emissions from Europe dominate BC and POA concentrations in Europe, with contributions in the range of 78 %–95 % and 58 %–78 %, respectively. The impact of local emissions on near-surface sulfate concentration is smaller than BC and POA. This is partially due to its less-efficient gas scavenging than the particles and the additional SO2-to-sulfate conversion process that increases the atmospheric residence time of sulfur. On the other hand, the higher initial injection height of SO2 emissions from some sources (e.g., industrial sector and power plants) facilitates the lifting of SO2 and sulfate aerosol into the free atmosphere and, therefore, favors long-range transport (Yang et al., 2019). The efficient reduction of local SO2 emissions in Europe also caused lower influences of local emissions on sulfate concentrations in recent years.

Anthropogenic emissions over oceans (e.g., international shipping) and natural emissions of oceanic dimethyl sulfide (DMS) and volcanic activities together account for 16 %–28 % of near-surface sulfate concentrations in the four subregions of Europe. About 10 % of sulfate and 5 %–10 % of BC and POA in E Europe and GTC come from Russia–Belarus–Ukraine emissions. Northern Africa contributes to 7 % of sulfate, 17 % of BC, and 24 % of POA over SW Europe. The contribution of emissions from the Middle East to aerosol concentrations in GTC are between 5 % and 10 %.

The transboundary and intercontinental transports of aerosols occur most frequently in the free troposphere rather than near the surface (Figs. S2 and S3). This also leads to larger relative contributions from non-European sources to aerosol column burdens than to the near-surface concentrations (Fig. 5). The European emissions only contribute 32 %–47 % of column burden of sulfate, 57 %–75 % of BC, and 51 %–71 % of POA over the four subregions of Europe. Over NW Europe and SW Europe, about 10 %–15 % of the sulfate burden is from East Asia and Russia–Belarus–Ukraine, respectively. Sources in northern Africa are responsible for 27 % and 14 % of BC and 19 % and 11 % of POA burden over SW Europe and GTC, respectively. Emissions from North America account for 15 % and 10 % of POA burden over NW Europe and SW Europe, respectively. Emissions from Russia–Belarus–Ukraine explain 12 % and 19 % of POA burden over E Europe and GTC, respectively. Since near-surface aerosol concentrations directly affect air quality and column burden is more relevant to radiative impact, the differences in relative contributions between near-surface concentrations and column burden highlight the possible roles of nonlocal emissions in either air quality or energy balance over Europe.

Source contributions to aerosols in Europe vary with season due to the seasonality of emissions and meteorology. In general, local sources have the largest contributions to both near-surface concentration and column burden of European aerosols in winter and the smallest contributions in summer (averaged over 2010–2018; outer rings in Fig. 6). With the contributions normalized by the ratio of the seasonal anthropogenic emission to the annual mean for each source, the impact of the seasonal variation in emissions on the source contributions can be removed (inner rings in Fig. 6) (Yang et al., 2019). Without the influence of emission seasonality, local source contributions decrease in winter and increase in summer, indicating that it was the higher local anthropogenic emissions that result in the larger local source contributions to wintertime aerosols in Europe relative to other seasons. Sulfur sources over oceans account for one-fourth to one-third of the European sulfate concentration and burden in spring likely due to the strong westerlies in this season that transport aerosols from the North Atlantic Ocean to the Europe. Source contributions from Russia–Belarus–Ukraine and North America to BC and POA in Europe show strong seasonal variabilities, which can be explained by the changes in biomass burning emissions considering their large seasonal variability.

Total sulfate-BC-POA concentrations decreased during 1980–2018 over all of the four subregions of Europe (Fig. 7) since near-surface aerosol concentrations in Europe are dominated by its local emissions and the European anthropogenic emissions have significantly decreased during this time period. Averaged over the entirety of Europe, near-surface concentrations of sulfate, BC, and POA decreased by 70 %, 43 %, and 23 %, respectively, between 1980–1984 and 2014–2018, which is consistent with the decreases in local emissions (Table 1). The total sulfate-BC-POA concentrations decreased by 62 %. With SOA included, this value does not have a substantial change (from 62 % to 59 %), and the decreasing trends in the four subregions of the Europe are not largely affected either. The column burden of sulfate, BC, POA, and the sum of these three decreased by 60 %, 28 %, 4 %, and 55 %, respectively, which are less than the decreases in corresponding near-surface concentrations. It is because nonlocal emissions have larger influences at high altitudes than at the surface, which partly dampened the contribution of near-surface aerosol decrease (induced by reductions in emissions) to the column integration.

The decrease in local European emissions explains 93 % of the reduced concentration and 91 % of the reduced burden in Europe between the first and last 5-year period of 1980–2018, while 8 %–9 % is contributed by the reduction in emissions from Russia–Belarus–Ukraine (Table 2). The decrease in emissions from North America also explains 10 % of the reduced column burden of sulfate-BC-POA in Europe from 1980–1984 to 2014–2018. Along with the decreases in local emission contributions to near-surface sulfate-BC-POA concentrations in Europe, the fraction of non-European emission contributions increased from 10 %–30 % to 30 %–50 % during 1980–2018 (Fig. 7), indicating that aerosols from foreign emissions have become increasingly important to air quality in Europe through long-range transport. Regulations for further improvement of air quality in Europe in the near future need to take changes in non-European emissions into account.

Similar to the declining trend in column burden, simulated total AOD also decreased from 0.12–0.16 to 0.06–0.08 in NW Europe and SW Europe and from 0.19–0.21 to 0.09–0.13 in E Europe and the GTC region during the past 4 decades (Fig. 8). Sulfate AOD accounts for the largest portion of total combustion AOD (sum of sulfate, BC, POA, and SOA) over the four subregions of Europe. The combustion AOD has decreased by 0.065 from 1980–1984 to 2014–2018 (Table 1), with 0.059 (91 %) contributed by the decrease in sulfate AOD. Therefore, we focus on sulfate aerosol when examining the decadal changes in AOD and DRF in Europe below.

The decreased sulfate AOD can also be decomposed into different contributions from individual source regions in CAM5-EAST. Local European emissions contribute to 89 % of the decrease, followed by 9 % and 7 % attributed to changes in emissions from Russia–Belarus–Ukraine and North America, respectively, with the residual offset by other source regions (Table 2). Over the last 4 decades, model-simulated sulfate AOD decreased at rates of 0.017, 0.017, 0.026, and 0.012 per decade, respectively, over NW Europe, SW Europe, E Europe, and GTC. Decreases in local European SO2 emissions result in 78 % of the sulfate AOD decreases over GTC and about 90 % over the other three subregions (Fig. 9). For the remote sources, emission changes in North America explain 5 %–10 % of the European sulfate AOD decrease, while Russia–Belarus–Ukraine sources contribute 29 % of the sulfate AOD decrease over GTC and 6 %–8 % over NW Europe and E Europe, indicating a possible warming enhancement effect of changes in emissions from North America and Russia–Belarus–Ukraine.

Averaged over 1980–2018, sulfate imposed a cooling effect over Europe, with the maximum negative DRF at the top of the atmosphere (TOA) exceeding -3 W m-2 in E Europe (Fig. 10). Compared to 1980–1984, the magnitude of sulfate DRF decreased in 2014–2018, leading to a 1–3 W m-2 warming mainly in E Europe. The warming effect mostly came from a local SO2 emission reduction, while non-European emission changes only contributed less than 0.4 W m-2 over most regions of the Europe. Considering Europe as a whole, the decrease in sulfate DRF caused a warming effect of 2.0 W m-2, with 88 % and 12 % coming from reductions in local European emissions and changes in non-European emissions, respectively (Tables 1 and 2).

Future changes in sulfate DRF associated with European and non-European emissions based on eight SSP scenarios are also estimated and shown in Fig. 11, and Fig. S4 gives an estimate for each SSP scenario. Sulfate DRF contributed by both European and non-European emissions would decrease in the near future but has large variabilities between different SSPs. The sulfate DRF (cooling) over Europe contributed from local European emissions shows a decrease from -0.48 W m-2 in the year 2015 to -0.18 (-0.08 to -0.33) W m-2 in the year 2030 and -0.14 (-0.05 to -0.29) W m-2 in the year 2050. Unlike their contributions to the historical (1980–2018) change, non-European emissions have an increasingly significant impact on the future sulfate DRF changes in Europe. The contributions of non-European emissions decrease from -0.68 W m-2 in the year 2015 to -0.39 (-0.13 to -0.64) W m-2 in the year 2030 and -0.26 (-0.08 to -0.63) W m-2 in the year 2050, with changes of a magnitude similar to that of local European emissions. It suggests that future changes in non-European emissions are as important as European emissions to radiative balance and associated regional climate change in Europe.

Using a global aerosol–climate model with an explicit aerosol source tagging technique (CAM5-EAST), we examine the long-term trends and source apportionment of aerosols in Europe for 1980–2018 from 16 source regions covering the globe in this study. CAM5-EAST can capture well the spatial distribution and temporal variation in aerosol species in Europe during this time period.

Averaged over 2010–2018, European emissions account for 54 %–68 %, 78 %–95 %, and 58 %–78 % of near-surface sulfate, BC, and POA concentrations over Europe, respectively. Russia–Belarus–Ukraine emissions explain 10 % of sulfate in E Europe and GTC. Northern Africa contributes to 17 % of BC and 24 % of POA over SW Europe. Anthropogenic emissions over oceans (e.g., from international shipping) and natural emissions from marine and volcanic activities together account for 16 %–28 % of sulfate near-surface concentrations in Europe. European emissions only account for 32 %–47 %, 57 %–75 %, and 51 %–71 % of column burden of sulfate, BC, and POA, respectively, in Europe, with the rest contributed by emissions from East Asia, Russia–Belarus–Ukraine, northern Africa and North America. Source contributions of aerosols in Europe vary with seasons, driven by the seasonality of emissions and meteorology.

Compared to 1980–1984, simulated total sulfate-BC-POA near-surface concentration and column burden for 2014–2018 had a decrease of 62 % and 55 %, respectively, the majority of which was contributed to by reductions in local European emissions. The decrease in emissions from Russia–Belarus–Ukraine contributed 8 %–9 % of the near-surface concentration decrease, while the decrease in emissions from North America accounted for 10 % of the reduced column burden. With the large decrease in local emission contributions, aerosols from foreign sources became increasingly important to air quality in Europe. The decrease in sulfate led to a 2.0 W m-2 warming in Europe, with 12 % coming from changes in non-European emissions, especially in North America and Russia–Belarus–Ukraine. Based on the SSP scenarios and the assumed relationship between DRF and emissions, we estimated that sulfate DRF over Europe that was contributed from European emissions and non-European emissions should decrease at a comparable rate in the near future. This suggests that future changes in non-European emissions are as important as European emissions in affecting regional climate change associated with aerosols in Europe. It should also be noted that the model currently does not have the ability to simulate nitrate and ammonium aerosols, and, therefore, the conclusions may not hold with all aerosols.