Modelling changes in secondary inorganic aerosol formation and nitrogen deposition in Europe from 2005 to 2030

Secondary inorganic PM_2.5 particles are formed from SO_x (${\mathrm{SO}}_{\mathrm{2}}+{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$), NO_x (NO+NO₂), and NH₃ emissions, through the formation of either ammonium sulfate ((NH₄)₂SO₄) or ammonium nitrate (NH₄NO₃). EU limits and WHO guidelines for PM_2.5 levels are frequently exceeded in Europe, in particular in the winter months. In addition the critical loads for eutrophication are exceeded in most of the European continent. Further reductions in NH₃ emissions and other PM precursors beyond the 2030 requirements could alleviate some of the health burden from fine particles and also reduce the deposition of nitrogen to vulnerable ecosystems.

Using the regional-scale EMEP/MSC-W model, we have studied the effects of year 2030 NH₃ emissions on PM_2.5 concentrations and depositions of nitrogen in Europe in light of present (2017), past (2005), and future (2030) conditions. Our calculations show that in Europe the formation of PM_2.5 from NH₃ to a large extent is limited by the ratio between the emissions of NH₃ on one hand and SO_x plus NO_x on the other hand. As the ratio of NH₃ to SO_x and NO_x is increasing, the potential to further curb PM_2.5 levels through reductions in NH₃ emissions is decreasing. Here we show that per gram of NH₃ emissions mitigated, the resulting reductions in PM_2.5 levels simulated using 2030 emissions are about a factor of 2.6 lower than when 2005 emissions are used. However, this ratio is lower in winter. Thus further reductions in the NH₃ emissions in winter may have similar potential to SO_x and NO_x in curbing PM_2.5 levels in this season.

Following the expected reductions of NH₃ emission, depositions of reduced nitrogen (${\mathrm{NH}}_{\mathrm{3}}+\mathrm{NH}{\mathrm{4}}^{+}$) should also decrease in Europe. However, as the reductions in NO_x emission are larger than for NH₃, the fraction of total nitrogen (reduced plus oxidised nitrogen) deposited as reduced nitrogen is increasing and may exceed 60 % in most of Europe by 2030. Thus the potential for future reductions in the exceedances of critical loads for eutrophication in Europe will mainly rely on the ability to reduce NH₃ emissions.

1 Introduction

Concentrations of particles with a diameter of less than 2.5 µm (PM_2.5) have been decreasing in most of Europe since the turn of the century as a combined result of reductions in anthropogenic emissions of primary particles and gaseous PM_2.5 precursors. Emissions of NH₃ play a central role in the secondary particle formation and are also major contributors to the exceedances of critical loads for eutrophication (Tsyro et al., 2020). In most parts of Europe emissions of in particular SO_x (emitted predominantly as SO₂ but also as ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$), and NO_x (NO+NO₂), have been steadily decreasing in the past decades. At the same time, emissions of NH₃ have changed much less, decreasing in some European countries and increasing in others (see Sect. 3 and Appendix B in EMEP Status Report 1/2020, 2020). Further reductions of SO_x, NO_x, and NH₃ emissions are required by the year 2030 according to the EU NEC2030 directive (https://www.eea.europa.eu/themes/air/air-pollution-sources-1/national-emission-ceilings, last access: 14 December 2021), but the projected percentage reductions in NH₃ emissions in NEC2030 are smaller than for SO_x and NO_x. In the atmosphere SO₂ is oxidised to ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and NO_x to HNO₃. Contrary to SO_x and NO_x, more than 90 % of the NH₃ emissions are from agriculture, with only minor contributions from industry and traffic (IIASA, 2020). As a result these emissions are in general not co-located with the SO_x and NO_x emissions. In addition the temporal distribution of the emissions differ, with NH₃ emissions peaking in spring and summer, whereas anthropogenic SO_x and NO_x emissions in general peak in winter. In the fine mode, ammonium sulfate ((NH₄)₂SO₄) particles are first formed from NH₃ and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$. Any excess NH₃ can then form ammonium nitrate (NH₄NO₃) in thermodynamic equilibrium with HNO₃ (see e.g. Simpson et al., 2012). When NH₃ is in excess relative to both ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations and the equilibrium with HNO₃, the formation of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ salts will slow down at some point (when there is less acid available to react with NH₃), and free NH₃ (NH₃ in excess of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$) will be present. With NH₃ emissions greatly exceeding SO_x and NO_x emissions already before 2005, one could question the effects of small or moderate reductions in NH₃ emissions on secondary inorganic aerosols (SIA25), a major component in PM_2.5.

Using emissions as described in Jiang et al. (2020), Aksoyoglu et al. (2020) showed that the fraction of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ in SIA25 was similar when calculated with 1990 versus 2030 emissions. With 1990 versus 2030 emissions, the fraction of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ in SIA25 dropped significantly, whereas the ${\mathrm{NO}}_{\mathrm{3}}^{-}$ fraction increased, compensating for the reduction in ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$. In many air pollution episodes in Europe involving PM_2.5, NH₄NO₃ has accounted for a large portion of the aerosol mass (Petit et al., 2017; Vieno et al., 2016). With a large surplus of NH₃ relative to HNO₃ it could be that NH₄NO₃ formation will be virtually unaffected by changes in NH₃ emissions. Both NO_x and NH₃ are relatively short-lived, with a lifetime in the atmosphere of about 1 d (Seinfeld and Pandis, 2016). Given the difference in both spatial and temporal distribution in the sources of NH₃, NO_x, and SO_x, substantial local variability in the ratio between NH₃ on one hand and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and/or HNO₃ on the other hand can be expected. Thus, locally the formation of SIA25 may be limited by the availability of either NH₃ or HNO₃ and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ due to the lack of co-location in both space and time of the sources of these species.

Here we apply the EMEP MSC-W model to investigate how PM_2.5 concentrations, and deposition of reduced nitrogen (${\mathrm{NH}}_{\mathrm{3}}+{\mathrm{NH}}_{\mathrm{4}}^{+}$), have changed from 2005 to 2017. But the main focus is on model calculations for 2030, assuming that the NEC2030 requirements will be met. Given that NH₃ concentrations in Europe are generally in substantial excess of HNO₃ concentrations, we explore to what extent additional reductions in NH₃ emissions will contribute to further reductions in ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and subsequently to reductions in PM_2.5 levels, and to what extent the response to further NH₃ emissions is linear. We try to answer this with a sensitivity study for PM_2.5 for post NEC2030, applying additional step-wise NH₃ emission reductions on top of the NEC2030 requirements holding all other emissions constant. At the same time we also investigate to what extent reductions in NH₃ emissions may affect deposition of reduced nitrogen and the exceedance of critical loads for nitrogen deposition.

2 Model description

The model calculations have been made with the EMEP MSC-W model (hereafter “EMEP model”), version rv4.34, on 0.1×0.1^∘ resolution for the domain between 30^∘ W and 45^∘ E and between 30 and 75^∘ N. A detailed description of the EMEP model can be found in Simpson et al. (2012), with later model updates described in Simpson et al. (2020) and references therein. In the EMEP model the composition of the metastable aqueous aerosols of the inorganic system containing ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and water and the system containing NH₃ and HNO₃ in the gas phase is calculated using the MARS equilibrium model (Binkowski and Shankar, 1995). In Tsyro and Metzgert (2019) the EMEP model results using the MARS model are compared to model calculations with EQUSAM4clim (Metzger et al., 2016, 2018), giving very similar results.

The EMEP model is available as open-source code (see code availability) and is under continuous development, receiving feedback from a host of users. It is regularly evaluated against measurements (see Gauss et al. (2017, 2018, 2019, 2020)) for the most recent evaluations. Scatter plots of model versus measurements for the concentrations of several key species, as well as for the wet depositions of reduced and oxidised nitrogen, are shown in Appendix A. The model performance is comparable for both 2005 and 2017, even though the selection of measurement sites differs for the two years. Measurements are also available for subsets of common sites for the two years, in general showing comparable biases between the model and measurements for the concentrations of key species for the years 2005 and 2017. The EMEP model has also participated in model intercomparisons and model evaluations in a number of peer-reviewed publications (Karl et al., 2019; Colette et al., 2011, 2012; Jonson et al., 2018). In Vivanco et al. (2018) depositions of sulfur and nitrogen species in Europe have been calculated by 14 regional models and compared to measurements, and in Theobald et al. (2019) the model-calculated trends in the wet deposition of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ as well as reduced and oxidised nitrogen from six models, including the EMEP model, are compared to measurements from 1990 to 2010. These two studies showed good results for the EMEP model. Out of the 14 models included in the study by Vivanco et al. (2018), the EMEP model was one of very few with low fractional biases compared to measurements for the wet depositions of reduced nitrogen (−0.01), oxidised nitrogen (−0.05), and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ (−0.11). For the trend studies presented in Theobald et al. (2019), the fractional bias for the years 1990 to 2010 was −0.18, −0.02, and 0.22 for the wet deposition of reduced nitrogen, oxidised nitrogen, and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ respectively, but the overall overestimation of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ was mainly caused by an overestimation in the first years of the period. Running the EMEP model in global mode, Ge et al. (2021) showed that the model captures the overall spatial and seasonal variations well for the major inorganic pollutants NH₃, NO₂, SO₂, HNO₃, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and wet depositions in East Asia, Southeast Asia, Europe, and North America.

Definition of the critical load for eutrophication

A critical load (CL) is defined as “a quantitative estimate of an exposure to one or more pollutants below which significant harmful effects on specified sensitive elements of the environment do not occur according to present knowledge” (Nilsson and Grennfelt, 1988). CLs are calculated for different receptors (e.g. terrestrial ecosystems, aquatic ecosystems), and a sensitive element can be any part (or the whole) of an ecosystem or ecosystem process. CLs have been derived for several pollutants and different negative effects. Here we restrict ourselves to CL defined to avoid the eutrophying effects of nitrogen deposition (CLeutN). Like sulfur, nitrogen can also have acidifying impacts in ecosystems, but the areas affected by acidification are strongly decreasing in Europe compared to earlier decades, and therefore the focus of this paper is on the eutrophying effects (Slootweg et al., 2015; EEA, 2014; Hettelingh et al., 2017).

The CLeutN for a site is either empirically derived or calculated from steady-state simple mass balance (SMB) equations. In the SMB method, non-harmful nitrogen-fixing processes are described mathematically and combined with a chemical criterion (e.g. an acceptable N concentration in the soil solution). This summation is then compared to the corresponding deposition value. Methods to compute CLs are summarised in the Mapping Manual of the ICP Modelling and Mapping CLRTAP (2017) (see also De Vries et al., 2015), which is used within the Convention on Long-range Transboundary Air Pollution (https://unece.org/40-years-clean-air, last access: 14 December 2021).

If the deposition of the pollutant under consideration is greater than the CL at a site, the CL is designated as exceeded. Such site-specific exceedances can be summarised for different spatial entities (e.g. grid cells, countries). This method is called average accumulated exceedance and is defined as the weighted average of exceedances for all ecosystems within the selected region, where the weights are the respective ecosystem areas (Posch et al., 2001).

The CL exceedances presented here were calculated using the current CL database, which is described in Hettelingh et al. (2017) and stored by the current Coordination Centre for Effects (CCE) at the German Federal Environmental Agency. The calculation is based on an extensive set of input data and equations. A detailed description is included in the Mapping Manual of the ICP Modelling and Mapping. (CLRTAP, 2017, Sect. 5). This dataset is also used, among other things, to support European assessments and negotiations on emission reductions (Hettelingh et al., 2001; Reis et al., 2012; EEA, 2014).

Figure 1 (a) Emissions of SO_x as SO₂, NO_x as NO₂, and NH₃ summed up for the EU28 countries. Percentage change in emissions from 2005 is given above the bars. (b) Concentrations of SIA (µg m⁻³) split into sulfate (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$), nitrate (${\mathrm{NO}}_{\mathrm{3}}^{-}$), and ammonium (${\mathrm{NH}}_{\mathrm{4}}^{+}$) and averaged over the EU28 countries. Percentage contributions to SIA are printed inside the bars.

3 Model runs

The EMEP model runs have been performed with 2017 meteorological conditions. In these model runs, emissions estimated for the years 2005 and 2017 and projected for 2030 have been used. The EMEP emissions are used as far as possible and supplemented by ECLIPSE (see below) emissions where needed. The model run with 2017 emissions is previously reported in the 2019 EMEP report (EMEP Status Report 1/2019, 2019). For the EU28 countries (EU28 includes the current EU27 countries and United Kingdom), the official EMEP emissions have been used for both the 2005 and 2017 model runs. For the model run with 2017 emissions, the EMEP emissions are also used for other countries and regions. The 2005 and 2017 EMEP emissions are listed in appendix B in the EMEP Status Report 1/2020 (2020). For the 2030 model runs the emissions for the individual EU28 countries are scaled from the 2005 emissions according to the NEC2030 obligations. The total emissions of NH₃, SO_x, and NO_x in the EU28 countries in 2005, 2017, and 2030 are illustrated in Fig. 1a. For all other countries and regions the 2005 and 2030 emissions have been provided by the International Institute for Applied Systems Analysis (IIASA) within the European FP7 project ECLIPSE (http://www.iiasa.ac.at/web/home/research/researchPrograms/air/ECLIPSEv5.html, last access: 14 December 2021). In this study we use ECLIPSE version 6a (hereafter referred to as “ECLIPSEv6a”), which is a global emission dataset widely used by the scientific community. Some of the methods used in ECLIPSEv6a are described in the recent publication of Höglund-Isaksson et al. (2020). NH₃ and NO_x emissions from all EU28 countries and selected non-EU European countries are listed in Table 1. For year 2005 non-EU28 EMEP and ECLIPSEv6a, emissions are very similar for NO_x, but for NH₃ the EMEP emissions are in general higher.

Table 1 Emissions of NO_x and NH₃, depositions of oxidised and reduced nitrogen, and the ratio of reduced to total (reduced + oxidised) deposition of nitrogen (Red fr). Emissions and deposition are listed in 100 Mg of N. Bosnia H. is Bosnia and Herzegovina, and N. Macedonia is North Macedonia. UK is the United Kingdom of Great Britain and Northern Ireland.

In order to explore the effects that further emission reductions of NH₃ in 2030 may have on PM_2.5 concentrations and nitrogen depositions, additional model sensitivity runs have been made. The 2030 NH₃ emissions have been reduced by up to 50 % in steps of 10 %. In addition the 2030 emissions of SO_x and NO_x have been reduced by 10 % and the 2005 NH₃ emissions by 10 %. All model runs are listed in Table 2.

Table 2 Model runs performed. All model runs have been performed with 2017 meteorological conditions as described in Sect. 3. Base denotes model runs with all emissions for the years 2005, 2017, and 2030. For 2005, emissions in the EU28 are based on EMEP 2005 official emissions. Remaining land-based emissions from ECLIPSEv6a. For 2017 all emissions are as reported in the EMEP Status Report 1/2020 (2020) Appendix B. For 2030 EU28 emissions are scaled according to the NEC2030 obligations based on the 2005 emissions. Remaining land-based emissions from ECLIPSEv6a. Emissions and model runs are also described in Sect. 3. The additional model sensitivity runs reducing the emissions in steps of 10 % are also listed.

Figure 2 PM_2.5 in 2005 (a) and in 2030 (b). Effects of 10 % further reductions in NH₃ emissions in 2005 (c) and in 2030 (d) (µg m⁻³).

4 Model results for 2005 versus 2030

4.1 PM_2.5

Figure 2 upper panels show concentrations of PM_2.5 as calculated with 2005 emissions (upper left) and 2030 emissions (upper right). The high PM_2.5 levels over North Africa in both 2005 and in 2030 are caused by large natural sources of mineral dust. As shown in Fig. 2, substantial reductions in PM_2.5 concentrations are expected from 2005 to 2030, caused by reductions in NH₃ emissions, and even larger reductions in SO_x and NO_x emissions, in Europe. Even so, in 2030 elevated PM_2.5 concentrations still persist in some areas, notably in the Po Valley in Italy and the Benelux countries (Belgium, the Netherlands, and Luxembourg). In these areas, anthropogenic primary PM_2.5 and PM_2.5 precursor emissions are expected to also remain high in 2030. As a result, the limit values for PM_2.5 recommended by WHO (WHO, 2005) are expected to also be exceeded in these locations in 2030.

Figure 1b shows the concentrations of SIA25 averaged over the EU28 countries in 2005, 2017, and 2030 split by component (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{NH}}_{\mathrm{4}}^{+}$). Even if the reductions in NH₃ emissions in the EU28 are much smaller than the corresponding reductions in SO_x and NO_x, the calculated percentage contributions to SIA25 from ${\mathrm{NH}}_{\mathrm{4}}^{+}$ are virtually unchanged between 2005, 2017, and 2030, confirming the findings in Aksoyoglu et al. (2020) for the Payerne measurement site in Switzerland. The lack of change in the fraction of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ is not surprising, as ${\mathrm{NH}}_{\mathrm{4}}^{+}$ is associated either with (NH₄)₂SO₄ or with NH₄NO₃. As the molecular weight is 18 g mol⁻¹ for NH₄, 96 g mol⁻¹ for ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, and 62 g mol⁻¹ for NO₃, the resulting percentage contribution by weight from NH₄ for both NH₄NO₃ and (NH₄)₂SO₄ is roughly 25 %, consistent with the contributions shown in Fig. 1b. Between 2005 and 2017 the percentage reductions in SO_x emissions in the EU28 were more than twice as large as the reductions in NO_x, resulting in an increase in the nitrate fraction in SIA25. From 2017 to 2030 the EU28 reductions in NO_x are expected to be larger than for SO_x, resulting in a slight decrease in the fraction of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and a corresponding increase in the ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ fraction in SIA25.

Table 3 Exceedance of CL for eutrophication (CLex eut.) by deposition (Dep.) of total nitrogen. Exceedance are expressed as share (%) of the receptor area.

Figure 3 Deposition of reduced N in 2005 (a) and in 2030 (b) (mg N m⁻² yr⁻¹).

4.2 Deposition of reduced nitrogen

Figure 3 shows the depositions of reduced nitrogen in 2005 (left) and in 2030 (right). As for PM_2.5, the Po Valley and the Benelux countries stand out, receiving large amounts of reduced nitrogen depositions both in 2005 and in 2030. The total amount of deposition of reduced nitrogen (and also oxidised nitrogen) per country in 2005, 2017, and 2030 is listed in Table 1. In most central European high-emitting countries less reduced nitrogen is deposited than is emitted. Several countries facing the sea, with very few upwind sources, exemplified by Ireland and Portugal, receive far less deposition than they emit. At the same time the Nordic countries (Norway, Sweden, and Finland) and the Baltic countries (Estonia, Latvia, and Lithuania), located downwind of central Europe, receive more deposition of reduced nitrogen than they emit. For the European Union as a whole, the fraction of deposited over emitted reduced nitrogen is between 0.7 and 0.8 for all three emission years considered. The remaining 0.2–0.3 is deposited either at sea or in non-EU countries. About 15 % of the NH₃ emitted within the model domain is advected out of the model domain, but much of this is coming from non-EU countries close the eastern model boundaries.

Figure 4 Fraction of reduced N deposition relative to total N (reduced plus oxidised nitrogen) deposition, calculated with year 2005, 2017, and 2030 emissions.

As a result of the lower ambitions for reductions in NH₃ emissions compared to NO_x emissions, a larger portion of the total nitrogen deposition is expected to come from NH₃. This is illustrated in Fig. 4, which shows the fraction of reduced nitrogen in the total nitrogen deposition calculated with 2005, 2017, and 2030 emissions. The figure shows that this fraction increases significantly from 2005 to 2017, with a further increase expected from 2017 to 2030. By 2030 the percentage of the total nitrogen deposition resulting from NH₃ emissions is expected to exceed 60 % in large parts of Europe. The percentage contributions are also listed as an average for EU28 and as averages for individual European countries in Table 1. This underpins the findings from IIASA (2018) that the potential for further reductions of the exceedances of CL for eutrophication mainly depends on our ability to control future NH₃ emissions. As shown in Fig. 5, the calculated CL for eutrophication is exceeded for all three years (2005, 2017, and 2030). Even though the level of exceedance has been substantially reduced from 2005 to 2017 and large reductions in depositions are also expected from 2017 to 2030, the total area in Europe where the CL is exceeded remains high for all three years. The percentage of the area where the CL for eutrophication is exceeded is listed in Table 3 for individual European countries.

Figure 5 Calculated exceedances of CL for eutrophication (eq ha⁻¹ yr⁻¹) in 2005, 2017, and 2030.

Figure 6 Fraction of NH₃ in reduced nitrogen (${\mathrm{NH}}_{\mathrm{3}}+{\mathrm{NH}}_{\mathrm{4}}^{+}$) in 2005, 2017, and 2030.

Table 4 First column listing annual and seasonal concentrations of PM_2.5 as an average for the EU28 countries. Columns 2–5 list the EU28 average reductions calculated in steps of 10 % reductions in NH₃ emissions. For PM_2.5 the reductions in ng N m⁻³ per gigagram of reduction of NH₃ emissions are shown in brackets. The corresponding effects of 10 and 20 % reductions of NH₃ emissions in 2005 are also shown. The effects of 10 % reductions of SO_x and NO_x emissions in 2030 are also listed. The reductions in PM_2.5 in ng N m⁻³ per gigagram emitted are given in brackets with NH₃ counted as NH₃ with molecular weight 17, NO_x counted as NO₂ with molecular weight 46, and SO_x counted as SO₂ with molecular weight 64.

4.3 Effects of NH₃ emission controls

Figure 2c and d show the effects of further 10 % reductions of NH₃ emissions on PM_2.5 concentrations in 2005 and 2030 respectively. Compared to 2005, the absolute effects of 10 % further emission reductions in 2030 are smaller. Partially, this is because the percentage emission reductions in 2005 give a larger reduction in absolute numbers compared to percentage emission reductions based on the lower 2030 emissions. As an example, 10 % of the emissions from the EU in 2005 (3574 Gg) will give a smaller reduction than 10 % reductions in 2030 (2900 Gg). However, these absolute changes in NH₃ emissions are not large enough to explain a decrease in the magnitude seen in Fig. 2c versus 2d. As seen in Fig. 6, there is more free NH₃, shown as NH₃ over total reduced nitrogen, in 2030 relative to 2005. A larger portion of free NH₃ could partially explain the 2 %–4 % annual increase in NH₃ observed by satellites between 2008 and 2018 in countries like Belgium, the Netherlands, France, Germany, Poland, Italy, and Spain (Damme et al., 2020). A 10 % reduction in NH₃ emissions will gradually a make smaller impact on the formation of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ in future years. This is exemplified by the EU28 countries in Table 4. For 2005, we find that as an annual average 10 % reductions of NH₃ emissions were about 4 times more efficient than 10 % reductions in NO_x and almost twice as efficient as SO_x in reducing PM_2.5 per gigagram emitted. For 2030, we find that as an annual average the efficiency of mitigating PM_2.5 concentrations by reducing NH₃ emissions by 10 % has been reduced from 0.61 to 0.22 ng N m⁻³ per Gg NH₃ emitted, a reduction of a factor of about 2.6 from 2005. Over the same time span, the efficiency of a further 10 % reduction in NO_x emissions has gone up by about a factor of 1.8 (from 0.15 to 0.27) and by about a factor of 1.6 (0.37 to 0.58) for a 10 % further reduction in SO_x emissions.

The dry deposition of NH₃ is faster than that of ${\mathrm{NH}}_{\mathrm{4}}^{+}$. As the fraction of NH₃ in total reduced nitrogen increases from 2005 to 2030 (as discussed in Sect. 4.1), reduced nitrogen may be deposited closer to its sources and potentially increasingly more in the same country as it is emitted. A trend in deposition versus emissions for the individual countries (deposition divided by emissions in Table 1) is not readily seen based on the model calculations. The geographical extent of the countries in Europe is relatively small, and there is considerable variability in the emission trends for NH₃ between the individual EU28 countries, also affecting the trends in the depositions in neighbouring countries.

As a large portion of the emitted reduced nitrogen is deposited close to its sources, changes in emissions close to the outer EU28 geographical borders should affect this fraction more for EU28 as a whole than emission changes in central parts. NH₃ emissions in large EU28 countries such as Germany and France have increased between 2005 and 2017, whereas emissions in several countries close to the eastern and southeastern geographical EU28 borders, such as Bulgaria, Romania, and Greece, have decreased. For the EU28 countries as a whole the fraction of deposited over emitted reduced nitrogen is between 0.7 and 0.8 for all three years considered (2005, 2017, and 2030). It would be possible to investigate the hypothesis of a possible decrease in the transport distance of reduced nitrogen by looking at so-called source receptor matrices for the different years (e.g. studying how the contribution from the country to itself have changed over the years). Such experiments are planned as a follow-up of this paper.

Figure 7 Effects of a 10 % decrease in 2030 NH₃ emissions on PM_2.5 (µg m⁻³) split by season. Winter: December, January, and February. Spring: March, April, and May. Summer: June, July, and August. Autumn: September, October, and November.

4.3.1 Seasonal effects of NH₃ emission reductions on PM_2.5

More than 90 % of the NH₃ emissions are from agriculture (IIASA, 2020), with low emissions in winter and a maximum in spring, as opposed to both NO_x and SO_x emissions peaking in winter. As a result, there is more ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and HNO₃ relative to NH₃ in winter than in other seasons. Also, the condensation process forming NH₄NO₃ aerosols is favoured by low temperatures. As a result Fig. 7 shows that for PM_2.5 by far the largest effects of further reductions of NH₃ emissions are modelled for the winter months. Notably, most of PM_2.5 pollution episodes, including exceedances of the EU limits or WHO AQ guidelines for daily mean PM_2.5 concentrations, are most frequent in large parts of Europe during the winter period (see Tsyro et al., 2019). The smallest effects are calculated for the summer months, when both SO_x and NO_x emissions are at a minimum. Thus in summer, reductions are mainly confined to the southwestern parts of the North Sea, where ship emissions of NO_x are large. This seasonal behaviour is also seen in the measurements at Preila in Lithuania, with low ${\mathrm{NH}}_{\mathrm{4}}^{+}$ concentrations in summer and higher concentrations in the cold season (Davuliene et al., 2021). Furthermore, they found that the relative abundance of NH₄NO₃ has increased at the expense of (NH₄)₂SO₄ as a result of particularly large reductions in SO_x emissions in the last decades.

The seasonal behaviour of PM_2.5 formation from NH₃ is also demonstrated for EU28 in Table 4, showing that the PM_2.5 reductions that can be achieved by reducing NH₃ emissions are largest in winter and are almost constant (and low) for each 10 % increment in emission reduction in summer. With a large surplus of free NH₃ in summer, the impact of further emission reductions is small. In winter the NH₃ surplus relative to HNO₃ and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ is much smaller (or nonexistent), and additional NH₃ emission reductions will have larger impacts on PM_2.5 levels.

Figure 8 Difference between additional 50 %–40 % reduction and additional Base–10 % in NH₃ emissions on top of NEC 2030 for PM_2.5 concentrations (µg m⁻³) (a) and deposition of reduced nitrogen (mg N m⁻²) (b).

4.3.2 Sensitivity tests with additional emission controls

Figure 8 compares the efficiency of NH₃ emissions reductions on top of the NEC2030 requirements for PM_2.5 concentrations and reduced nitrogen depositions. Starting from the expected emission levels in 2030, the maps compare the effects of the first 10 % reductions (Base–10 %) in NH₃ emissions to the effects of further reductions in NH₃ emissions from 40 %–50 % relative to Base. If linear, the effects of these 10 % increments in emissions should be equal. However, as shown in Fig. 8a, the reductions in PM_2.5 are larger for the 50 %–40 % emission reductions compared to Base–10 % reductions almost everywhere. This is further demonstrated in Table 4, listing the reductions in annual and seasonal PM_2.5 concentrations as an average over the EU28 countries in steps of 10 % relative to the 2030 NEC emissions. Both as an annual average and for each season, the reductions in PM_2.5 increase for each 10 % increment. The reductions in PM_2.5 increase from 0.23 ng m⁻³ per gigagram of NH₃ emitted for the first 10 % additional reductions to 0.35 ng m⁻³ per gigagram emitted for the 50 %–40 % reductions. The increase in efficiency is a result of a shift in the ratio in NH₃ versus SO_x and NO_x emissions. In Table 4 we also show that with the much higher SO_x and NO_x emissions versus NH₃ emissions, the potential of 10 % additional reductions in NH₃ emissions in curbing PM_2.5 levels was substantially higher in 2005, even when compared to 40 %–50 % reductions in 2030.

In two additional model runs we separately reduce the 2030 emissions of SO_x and NO_x by 10 %. As an annual average we find that 10 % reductions in both SO_x and NO_x emissions will lead to larger reductions in PM_2.5 levels in the EU28 than the corresponding 10 % reductions in NH₃ emissions. However, as discussed in Sect. 4.3.1, the seasonal variation is large, and in winter reductions of PM_2.5 per gigagram emitted could still be larger for NH₃ than for NO_x.

For depositions of reduced nitrogen the situation is reversed. As shown in Fig. 8b the reductions in depositions achieved with NH₃ reduced between 50 % and 40 % compared to the first 10 % reductions are (marginally) smaller in the vicinity of the source regions. This can be explained by a slightly larger portion of the emitted NH₃ being converted to ${\mathrm{NH}}_{\mathrm{4}}^{+}$ aerosols and having a slower dry deposition rate than NH₃. As a result, the higher deposition seen in the source areas is compensated for by a much smaller but more widespread decrease elsewhere. As discussed in Sect. 4.2, only a small portion of the reduced nitrogen is advected out of the central parts of the model domain.

5 Discussion and conclusions

Focusing on the effects of NH₃ emissions, we have investigated how PM_2.5 concentrations and depositions of reduced nitrogen will change from 2005 to 2030, assuming that the NEC2030 emission targets will be met. In addition, we have made a sensitivity study for PM_2.5 for post NEC2030, assuming additional emission reductions on top of the NEC2030 requirements.

Emissions of SO_x and NO_x have decreased in Europe from the year 2005 to present, and further emissions reductions are expected by the year 2030. However, NH₃ emissions have so far remained high, and projected NEC2030 emission reductions of NH₃ are much smaller than for SO_x and NO_x. Our model calculations show that these differences in emission trends lead to a smaller fraction of the emitted NH₃ being converted to ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and an increasingly larger portion of free NH₃ versus ${\mathrm{NH}}_{\mathrm{4}}^{+}$ in the atmosphere in Europe. Based on 10 % emission reductions of NH₃, NO_x, and SO_x, we calculate that the potential for PM_2.5 formation per gigagram of NH₃ emitted is expected to drop by a factor of about 2.6 as an annual average between 2005 and 2030. Over the same time span the potential for forming PM_2.5 from NO_x per gigagram emitted has increased by a factor of 1.8 and from SO_x by a factor of 1.6 per gigagram emitted.

In winter, with low NH₃ emissions and relatively higher NO_x and SO_x emissions, the ratio of NH₃ to HNO₃ and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ is higher, and a larger portion of the emitted NH₃ will form particulate ${\mathrm{NH}}_{\mathrm{4}}^{+}$. Also, the formation of NH₄NO₃ in equilibrium with HNO₃ and NH₃ is favoured by low temperatures. As a result we find that in winter the effects of further reductions in NH₃ emissions are larger than in other seasons and comparable to additional reductions in SO_x and NO_x emissions. This is in agreement with the findings in Backes et al. (2016), pointing out that even though the NH₃ emissions are highest in spring and summer due to the application of manure on the fields, emission reductions in winter have a stronger impact on the formation of secondary aerosols than in any other season. Furthermore they stated that the potential of reducing NH₃ emissions in winter is highest through the reduction of animal farming, as this source accounts for about 80 % of the NH₃ emissions in the autumn and winter months.

Following the emission reductions of NH₃, deposition of reduced nitrogen is decreasing in Europe. However, the reductions in NO_x emissions are much larger than for NH₃, resulting in a much faster decline in oxidised nitrogen deposition compared to reduced nitrogen. Thus the fraction of reduced over total deposition of nitrogen is increasing and is expected to reach more than 60 % in large parts of Europe by the year 2030. Our calculations show that with the existing emission projections the CL for nitrogen will also be exceeded in large parts of Europe in 2030. There are also indications that reduced nitrogen inputs are more effective in decreasing biodiversity than oxidised nitrogen, as reduced nitrogen is more readily available, stimulating growth of specific plants at the expense others (see van den Berg et al., 2008, and Erisman et al., 2007, and references therein). Furthermore they also suggest that increased levels of ${\mathrm{NH}}_{\mathrm{4}}^{+}$ can be toxic to plants (see also Esteban et al., 2016).

Reducing, and preferably removing, these exceedances will require larger reduction in nitrogen emissions than currently projected. Given that reduced nitrogen is responsible for the major fraction of nitrogen depositions, the largest cuts should be made in the NH₃ emissions.

As discussed in Nenes et al. (2020, 2021), gas–aerosol partitioning of total reduced and oxidised nitrogen is affected by aerosol pH level and water content, so that low (high) pH is favourable for ${\mathrm{NH}}_{\mathrm{4}}^{+}$ (${\mathrm{NO}}_{\mathrm{3}}^{-}$) formation. The increase in the aerosol fraction in total reduced and oxidised nitrogen would lead to changes in their dry deposition and subsequently their residence times and transport distances. This effect has not been accounted for in the EMEP model. Thus some limited local effects might have been missed in our model simulations. For instance, based on the Nenes et al. (2021) results, there may be additional ${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation in areas with low acidity, such as coastal or dusty regions. Potentially this may reduce the deposition of total nitrate near these local sources, somewhat enhancing the accumulation of particles. Furthermore, as future emissions of SO_x and NO_x are expected to decrease, the pH of the particles is likely to increase, potentially favouring ${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation and thus decreasing dry deposition and increasing the transport distances of oxidised and thereby total nitrogen in some regions. On the other hand, our results show that overall, the fraction of reduced nitrogen in the total nitrogen has been increasing, and this increase is expected to continue until 2030. Assuming that the deposition rates for total nitrogen are mostly driven by those of reduced nitrogen (following Nenes et al., 2021), the local effects of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ formation bursts would probably not play a major role across the regions in different present and future chemical regimes. Therefore we believe that overall the main conclusions presented in our paper remain valid.

For many countries the latest source-oriented legislation may potentially reduce the emissions of SO_x and NO_x below their emission reduction requirements, and as a result the EU28 as a whole could be on track to overshoot the reduction requirements for these species by 2030. But for NH₃ further efforts are needed in order to meet the 2030 commitments for many countries in Europe (IIASA, 2018). Cost-effective measures to further reduce NH₃ emissions differ among various parts of Europe. According to IIASA (2020) the damage cost estimate of EUR 17.50 per kilogram of NH₃ emitted is much higher than the average abatement costs. Also, Giannakis et al. (2019) find that much more ambitious commitments for NH₃ emission reductions could be applied by EU28 countries with relatively minimal costs. According to Giannakis et al. (2019), low-emission animal housing would be the least cost-effective measure, but still a quarter of the costs of the avoided damage. However, this measure could be the most effective one to reduce emissions in winter, when PM_2.5 health-related limits are the most likely to be exceeded in Europe.

Appendix A

In this appendix we show scatter plots of EMEP model results versus measurements for some key species for the years 2005 and 2017. Several of the 2017 scatter plots have already been published in Gauss et al. (2019) but are shown here in order to provide a direct comparison to the year 2005 results. For the year 2017 the comparison to measurements differs slightly from the results shown in Gauss et al. (2019) as several new measurements have been made available since the publication of the above-mentioned report. For both years the scatter plots are shown with all measurements included. However, the selection of available measurements differs between the two years. In general there are more measurements available in 2017 compared to 2005. In Table A1 statistics for the scatter plots are listed for the same species as in the scatter plots, limiting the sites to those having measurements for both 2005 and 2017. The number of common measurements for both years is given in the table. In particular for NH₃ in air and HNO₃ in air there are very few sites with measurements for both years.

Figure A1 Scatter plots of modelled versus observed concentrations of NH₃ (a, b) and ammonium (c, d) in air for the years 2005 (a, c) and 2017 (b, d). Statistics with a limited set of common measurement sites are given in Table A1. Site positions are listed in Table A2. Measurements were downloaded from http://ebas.nilu.no (last access: 14 December 2021).

Figure A2 Scatter plots of modelled versus observed concentrations of nitric acid (a, b) and nitrate (c, d) in air for the years 2005 (a, c) and 2017 (b, d). Statistics with a limited set of common measurement sites are given in Table A1. Site positions are listed in Table A2. Measurements were downloaded from http://ebas.nilu.no (last access: 14 December 2021).

Figure A3 Scatter plots of modelled versus observed wet depositions of reduced nitrogen (a, b) and oxidised nitrogen (c, d) for the years 2005 (a, c) and 2017 (b, d). Statistics with a limited set of common measurement sites are given in Table A1. Site positions are listed in Table A2. Measurements were downloaded from http://ebas.nilu.no (last access: 14 December 2021).

Figure A4 Scatter plots of modelled versus observed concentrations of sulfate in air (a, b) and wet deposition of sulfate (c, d) for the years 2005 (a, c) and 2017 (b, d). Statistics with a limited set of common measurement sites are given in Table A1. Site positions are listed in Table A2. Measurements were downloaded from http://ebas.nilu.no (last access: 14 December 2021).

Table A1 Year 2005 and year 2017 statistics for scatter plots with common measurement sites for both years. Obs is observed annual mean for the sites. Bias is model bias, and Corr. is correlation between measurements and the model calculations.

Table A2 List of sites included in scatter plots in Figs. A1 to A4. Wet depositions when included for all three species (reduced nitrogen, oxidised nitrogen, and oxidised sulfur) are marked with ✓. Only reduced and oxidised nitrogen marked with ∇, only oxidised N and S marked with △, only reduced N and oxidised S marked with •, and only oxidised sulfur marked with ⋄.