Global sensitivities of reactive N and S gas and particle concentrations and deposition to precursor emissions reductions

The reduction of fine particles (PM_2.5) and reactive N (N_r) and S (S_r) species is a key objective for air pollution control policies because of their major adverse effects on human health, ecosystem diversity, and climate. The sensitivity of global and regional N_r, S_r, and PM_2.5 to 20 % and 40 % individual and collective reductions in anthropogenic emissions of NH₃, NO_x, and SO_x (with respect to a 2015 baseline) is investigated using the EMEP MSC-W (European Monitoring and Evaluation Programme Meteorological Synthesizing Centre – West) atmospheric chemistry transport model with WRF (Weather Research and Forecasting) meteorology. Regional comparisons reveal that the individual emissions reduction has multiple co-benefits and small disbenefits on different species, and those effects are highly geographically variable. A 40 % NH₃ emission reduction decreases regional average NH₃ concentrations by 47 %–49 % but only decreases NH${}_{\mathrm{4}}^{+}$ by 18 % in Euro_Medi, 15 % in East Asia, 12 % in North America, and 4 % in South Asia. This order follows the regional ammonia richness. A disbenefit is the increased SO₂ concentrations in these regions (10 %–16 % for 40 % reductions) because reduced NH₃ levels decrease SO₂ deposition through altering atmospheric acidity. A 40 % NO_x emission reduction reduces NO_x concentrations in East Asia by 45 %, Euro_Medi and North America by ∼ 38 %, and South Asia by 22 %, whilst the regional order is reversed for fine NO${}_{\mathrm{3}}^{-}$, which is related to enhanced O₃ levels in East Asia (and also, but by less, in Euro_Medi) and decreased O₃ levels in South Asia (and also, but by less, in North America). Consequently, the oxidation of NO_x to NO${}_{\mathrm{3}}^{-}$ and of SO₂ to SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is enhanced in East Asia but decreased in South Asia, which causes a less effective decrease in NO${}_{\mathrm{3}}^{-}$ and even an increase in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in East Asia but quite the opposite in South Asia. For regional policy making, it is thus vital to reduce three precursors together to minimize such adverse effects. A 40 % SO_x emission reduction is slightly more effective in reducing SO₂ (42 %–45 %) than SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (34 %–38 %), whilst the disbenefit is that it yields a ∼ 12 % increase in total NH₃ deposition in the four regions, which further threatens ecosystem diversity. This work also highlights important messages for policy makers concerning the mitigation of PM_2.5. More emissions controls focusing on NH₃ and NO_x are necessary for regions with better air quality, such as northern Europe and eastern North America. In East Asia, the three individual reductions are equally effective, whilst in South Asia only SO_x reduction is currently effective. The geographically varying non-one-to-one proportionality of chemical responses of N_r, S_r, and PM_2.5 to emissions reductions revealed by this work show the importance of both prioritizing emissions strategies in different regions and combining several precursor reductions together to maximize the policy effectiveness.

1 Introduction

Reactive N (N_r) and S (S_r) species are critical determinants of air quality. A substantial proportion of ambient PM_2.5 (particulate matter with aerodynamic diameter ≤ 2.5 µm) is secondary inorganic aerosol (SIA) formed from chemical reactions of emissions of the precursor gases NH₃, NO_x (NO and NO₂), and SO_x (sulfur oxides, mainly SO₂) (Behera et al., 2013; Weber et al., 2016; Vasilakos et al., 2018; Nenes et al., 2020). PM_2.5 is consistently associated with elevated risk of all-cause mortality and other adverse health impacts (Hart et al., 2015; Chen et al., 2017; C. Chen et al., 2018; Karimi et al., 2019; Stieb et al., 2020). The gases NO₂ and SO₂ are also direct health pollutants. In addition, oxidized N (e.g. NO_x, HNO₃, and NO${}_{\mathrm{3}}^{-}$, collectively abbreviated as OXN) and reduced N (e.g. NH₃ and NH${}_{\mathrm{4}}^{+}$, collectively abbreviated as RDN) species are powerful nutrients for plants and microorganisms, whose deposition leads to eutrophication and loss of ecosystem biodiversity (Erisman et al., 2005; Bergström and Jansson, 2006; Sun et al., 2017; Kharol et al., 2018). The severity of the adverse effects of N deposition is determined not only by the total quantity but also by its form. Many studies show different N deposition components have varying toxicity to different plants; dry deposition of NH₃ is particularly deleterious, for example (Van Herk et al., 2003; Sheppard et al., 2011; Sutton et al., 2014, 2020; Pescott et al., 2015). Deposition of oxidized S (i.e. SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$, collectively abbreviated as OXS) also greatly influences precipitation acidity (Lu et al., 2010; Aas et al., 2019; McHale et al., 2021). These observations put greater emphasis on mitigation of certain deposition components.

The East Asia, South Asia, Euro_Medi (Europe and Mediterranean), and North America regions have high population density and high N_r and S_r pollution. Historically, Europe and North America were the dominant emissions regions, suffering severe air pollution until the late 20th century. As reductions in SO_x and NO_x emissions took effect in Europe and North America, emissions in East and South Asia increased dramatically due to rapid industrialization and dominated global N_r and S_r emissions by the early 21st century (Weber et al., 2016; NEC, 2019; Fowler et al., 2020), although China, in particular, is now implementing effective SO_x and NO_x emissions controls (Liu et al., 2016; Hoesly et al., 2018; Zheng et al., 2018; Meng et al., 2022). In contrast, a lack of action on NH₃ emissions in most countries, coupled with the growth in agriculture to feed a rising global population, means that global NH₃ emissions continue to grow (Heald et al., 2012; Fowler et al., 2015; Aksoyoglu et al., 2020). As a result, ambient N_r and S_r pollution remains a major health and environmental concern in most regions. The European Environmental Agency reported that 97 % of the urban population in the European Union in 2019 was exposed to annual mean concentrations of PM_2.5 above the latest World Health Organization (WHO) air quality guideline of 5 µg m⁻³ (EEA, 2021; WHO, 2021), whilst the United States Environmental Protection Agency reported that for the period 2014–2016 only 10 % of its 429 monitoring sites had PM_2.5 concentrations < 6.0 µg m⁻³ (USEPA, 2017). Combining satellite retrievals, chemistry model simulations, and ground level measurements, Ma et al. (2014) and Brauer et al. (2016) showed that in 2013 the majority of the East and South Asia populations lived in areas where annual mean PM_2.5 concentrations exceeded the WHO Interim Target 1 of 35 µg m⁻³ (WHO, 2021). Furthermore, as mitigation of NO_x emissions in recent years has been more effective than for NH₃, deposition of RDN is now increasingly responsible for the exceedances of N critical loads for eutrophication in many regions (Jovan et al., 2012; X. Chen et al., 2018; Simpson et al., 2020a; Yi et al., 2021; Jonson et al., 2022).

Understanding the sensitivities of PM_2.5, N_r, and S_r pollution to emissions reductions is complicated not only by the substantial regional heterogeneity in relative emissions but also by the substantial meteorological heterogeneity influencing the chemistry and deposition. This necessitates use of atmospheric chemistry transport models (ACTMs) designed to simulate the underlying physical–chemical processes linking emissions, dispersion, chemical reactions, and deposition of atmospheric components. Previous ACTM studies have provided insight into the complexities of the sensitivity of PM_2.5 and its SIA components to changes in emissions in different regions that measurements cannot reveal. For example, using the GEOS-Chem model, Wang et al. (2013) showed that SIA concentrations in 2015 decreased in South China and the Sichuan Basin but increased in North China compared to their 2006 levels in response to −16 % SO₂ and +16 % NO_x emissions changes (no change in NH₃ emissions) from 2006 to 2015 according to China's 12th 5-year plan, but if NH₃ emissions increase by +16 % (based on their growth rate from 2006 to 2015), the SIA reduction due to SO₂ reduction will be totally offset in all regions because of the elevated NH₃NO₃ formation, demonstrating the importance of NH₃ control on China's SIA mitigation. Pommier et al. (2018) reported substantial projected growth in emissions in India between 2011 and 2050, amounting to 304 % for SO_x, 287 % for NMVOCs (non-methane volatile organic compounds), 162 % for NO_x, 100 % for primary PM_2.5, and 60 % for CO and NH₃, leading to increases in annual mean PM_2.5 and O₃ concentrations of 67 % and 13 %, respectively. In the UK, results from EMEP4UK model simulations for 2010 emissions and meteorology indicated that NH₃ emissions reductions are the most effective single-component control (compared to individual reductions in NO_x, SO_x, and primary PM_2.5) on area-weighted PM_2.5, whilst weighting by population placed greater emphasis on reductions in emissions of primary PM_2.5 (Vieno et al., 2016). Holt et al. (2015) used GEOS-Chem to investigate PM_2.5 sensitivities in the United States to emissions reductions between two sets of scenarios representing a 2005 baseline (high emissions) and a 2012 analogue (low emissions). They found larger sensitivities of PM_2.5 to SO_x and NO_x controls in the low emissions case, since lower NO_x emissions in 2012 enhance the relative importance of aqueous-phase SO₂ oxidation.

These studies analysed N_r and S_r responses to precursor emissions reductions in the early 2000s and in specific regions but do not provide a global view of sensitivities to the same reductions everywhere. Given the considerable emissions changes in global and regional NH₃, NO_x, and SO_x in recent years (Hoesly et al., 2018; Kurokawa and Ohara, 2020), our understanding of the current chemical climate for N_r and S_r reactions on the global and regional scale and how it affects responses of PM_2.5, N_r and S_r species to various emissions reductions should be updated. This is the motivation for the work presented here, which provides a global picture of the effectiveness of NH₃, NO_x, and SO_x emissions reductions for mitigating both concentration and deposition of N_r and S_r pollutants. We used the EMEP MSC-W (European Monitoring and Evaluation Programme Meteorological Synthesizing Centre – West) ACTM to simulate the global domain based on global emissions and meteorology in 2015, which enables a regional comparison to be conducted with inherently consistent simulations. The focus here is on annual means, as these are the long-term metric within global and regional air quality standards. We first describe the model set-up and performance and the sensitivity experiments used to simulate responses of PM_2.5, N_r, and S_r species to 20 % and 40 % reductions in gaseous precursor emissions (Sect. 2). Section 3 details the global and regional concentration and deposition changes in components of RDN, OXN, OXS, and PM_2.5 between baseline and emissions reduction scenarios. Section 4 discusses key processes that determine the benefits and disbenefits of emissions reductions and how they vary geographically, as well as the implications of our findings for policy making.

2 Methods

2.1 Model set-up and performance

The EMEP MSC-W open-source atmospheric chemistry transport model (https://www.emep.int, last access: 8 August 2022) is a three-dimensional Eulerian model, widely used for both scientific research and policy development (Bergström et al., 2014; Jonson et al., 2017; Pommier et al., 2018; McFiggans et al., 2019; Karl et al., 2019; Pommier et al., 2020; Jonson et al., 2022). Version rv4.34 was used here. A detailed technical description of EMEP MSC-W rv4.0 is documented in Simpson et al. (2012). A series of overviews of model updates from version rv4.0 to rv4.34 is documented in annual EMEP status reports (Simpson et al., 2013; Tsyro et al., 2014; Simpson et al., 2015, 2016, 2017, 2018, 2019, 2020b). Meteorology for 2015 was derived from the Weather Research and Forecasting model (WRF, https://www.mmm.ucar.edu/models/wrf; https://github.com/wrf-model/WRF/releases/tag/v4.2.2, last access: 8 August 2022) version 4.2.2. The coupled EMEP–WRF system has been tested and applied to many regional and global studies (Vieno et al., 2010, 2014, 2016; Werner et al., 2018; Chang et al., 2020; Gu et al., 2021).

Detailed global EMEP–WRF configurations used in this work are presented in Y. Ge et al. (2021, 2022). In brief, the global domain has a horizontal resolution of 1^∘ × 1^∘ and 21 terrain-following vertical layers from the surface up to 100 hPa. The height of the lowest model layer is around 45 m. The model outputs of surface concentrations are adjusted to correspond to 3 m above ground level in order to provide concentrations at heights more typical of ambient measurements and human exposure (Simpson et al., 2012). The aerosol module is the EQuilibrium Simplified Aerosol Model V4 (EQSAM4clim), which parameterizes a full gas–liquid–solid partitioning scheme for semi-volatile and non-volatile mixtures. Details are described in Metzger et al. (2016, 2018). Dry deposition of gaseous species and aerosol components to the ground surface is simulated, utilizing deposition velocity as described in Simpson et al. (2012, 2020a). The parameterization of wet deposition incorporates both in-cloud and below-cloud scavenging of gases and particles (Berge and Jakobsen, 1998; Simpson et al., 2012).

The global model evaluation of N_r and S_r concentrations and wet deposition from this model configuration for 2010 and 2015 against measurements from 10 ambient monitoring networks is documented in Y. Ge et al. (2021) and demonstrates the model's capability for capturing the spatial and seasonal variations of NH₃, NH${}_{\mathrm{4}}^{+}$, NO₂, HNO₃, NO${}_{\mathrm{3}}^{-}$, SO₂, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in East Asia, South East Asia, Europe, and North America. For instance, the correlation coefficients between global model and measurement annual mean concentrations for most species in 2015 are ≥0.78 and for annual wet deposition of RDN and OXN are 0.78 and 0.63, respectively. This is in spite of inherent uncertainty in both model and measurements and differences in their spatial representativeness. Section S1 in the Supplement also gives a brief introduction of the model performance compared to measurements for RDN species, as an example.

2.2 Emissions and model experiments

Baseline emissions for 2015 were from the ECLIPSE V6 (Evaluating the Climate and Air Quality Impacts of Short-Lived Pollutants) inventory, available at https://previous.iiasa.ac.at/web/home/research/researchPrograms/air/ECLIPSEv6b.html (last access: 8 August 2022). Monthly emissions profiles derived from EDGAR (Emission Database for Global Atmospheric Research, v4.3.2 datasets, available at https://edgar.jrc.ec.europa.eu/dataset_temp_profile, last access: 31 May 2023) time series (Crippa et al., 2020) were applied to the ECLIPSE annual emissions of SO₂, NO₂, NH₃, CO, CH₄, NMVOC, primary PM_2.5, and coarse particles. Detailed implementation procedures including the re-assignment of ECLIPSE emissions sectors to EMEP sectors and the calculation of temporal profiles at a given country for a given pollutant are described in Y. Ge et al. (2021).

A baseline simulation and a set of eight sensitivity experiments were conducted for emissions and meteorology for 2015. Limited by available computational resources and storage space and taking the achievability of real-world emissions controls into account, the model experiments applied 20 % and 40 % reductions to global anthropogenic emissions of NH₃, NO_x, and SO_x from all sectors both individually and collectively (i.e. reductions applied to all three species simultaneously). All other emissions, including natural emissions such as dimethyl sulfide from oceans, lightning NO_x,, and soil NO_x, were left unchanged.

The sensitivity (Sensitivity_i) of the concentration/deposition of a species i is calculated as the absolute difference between the value in baseline (Baseline_i) and in an emission reduction scenario (Scenario_i). Taking NH₃ concentration as an example

$$\begin{array}{}\text{(1)}& {\mathrm{Sensitivity}}_{{\mathrm{NH}}_{\mathrm{3}}}\left(\mathrm{\mu }\mathrm{g}{\mathrm{m}}^{-\mathrm{3}}\right)={\text{Baseline}}_{{\mathrm{NH}}_{\mathrm{3}}}-{\text{Scenario}}_{{\mathrm{NH}}_{\mathrm{3}}}.\end{array}$$

For the relative sensitivity (Relative Sensitivity_i)

$$\begin{array}{}\text{(2)}& {\text{Relative Sensitivity}}_{{\mathrm{NH}}_{\mathrm{3}}}\left(\mathit{\%}\right)={\displaystyle \frac{{\text{Sensitivity}}_{{\mathrm{NH}}_{\mathrm{3}}}}{{\text{Baseline}}_{{\mathrm{NH}}_{\mathrm{3}}}}}\times \mathrm{100}\mathit{\%}.\end{array}$$

The sensitivities of different species are calculated for all emissions reduction scenarios. The PM_2.5 sensitivities derived from individual reductions in emissions of NH₃, NO_x, or SO_x are used to define the sensitivity regimes for different regions in Sect. 3.2. For each model grid, the regime is decided by the precursor that yields the greatest decrease in grid PM_2.5 concentration: NH₃ sensitive, NO_x sensitive, or SO_x sensitive.

2.3 Definition of world regions

We compared the sensitivities to the emissions reductions of PM_2.5, N_r, and S_r species concentrations and depositions in the four world regions of East Asia, South Asia, Euro_Medi, and North America, defined in Fig. 1 (and listed in Table S1 in the Supplement). These are based on regions used by the Intergovernmental Panel on Climate Change and as rationalized in Iturbide et al. (2020). All four regions are densely populated and have high N_r and S_r pollution. Besides, due to limitations in the number of publicly available measurements, our model outputs are evaluated against measurements in East Asia, Europe, and North America, and therefore we have greater confidence in sensitivity results in these three regions. South Asia is chosen because of its extreme ammonia richness, as revealed by Ge et al. (2022), which makes it an interesting comparison with other regions.

Figure 1 The boundaries of the four world regions used in this study, which are based on the IPCC reference regions described in Iturbide et al. (2020).

3 Results

3.1 Sensitivities of N_r and S_r gas and aerosol concentrations

The simulated global baseline 2015 annual mean surface concentrations of N_r and S_r have been discussed in detail in Ge et al. (2022). Here we analyse the sensitivities of the modelled surface concentrations to SIA precursor emissions reductions for RDN, OXN, and OXS components. Sensitivities differ according to consideration of primary or secondary components and show great geographical variation.

3.1.1 RDN

Figure 2 shows the spatial variations in the sensitivities of NH₃ and NH${}_{\mathrm{4}}^{+}$ annual mean surface concentrations to 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Regional average sensitivities in East Asia, South Asia, Euro_Medi, and North America are summarized in Fig. 3 and Table S2. Steeper gradients in Fig. 3 correspond to greater concentration changes (sensitivities).

Figure 2 Changes in NH₃ and NH${}_{\mathrm{4}}^{+}$ annual surface concentrations for 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

From these figures, it is clear that whilst reducing emissions of NH₃ (and all three precursors together) decrease NH₃ concentrations efficiently, reducing emissions of NO_x or SO_x lead to increases in NH₃ concentrations, particularly over densely populated areas. The maximum reduction in model grid NH₃ concentration across all scenarios reaches 16.6 µg m⁻³ (44 % change relative to baseline; expressed similarly hereafter) and occurs over East Asia in response to a 40 % reduction in NH₃ emissions. The largest increase in NH₃ concentration (1.51 µg m⁻³, 22 %) arises in South East Asia under the 40 % SO_x emissions reduction scenario. Lower NO_x and SO_x emissions decrease the concentration of acidic species available to react with NH₃ to form NH${}_{\mathrm{4}}^{+}$ aerosol, so more of the emitted NH₃ stays in the gas phase. Other studies have likewise shown that reductions in SO₂ or NO_x emissions are an important contributor to the growth in tropospheric NH₃ concentrations globally and regionally (Saylor et al., 2015; Warner et al., 2017; Liu et al., 2018; Yu et al., 2018).

Figure 3 The absolute and relative sensitivities of regionally averaged annual mean surface concentrations of NH₃ (upper row) and NH${}_{\mathrm{4}}^{+}$ (lower row) to 20 % and 40 % emissions reductions in NH₃ (blue), NO_x (orange), and SO_x (green), individually and collectively (red), for the four regions defined in Fig. 1. The solid grey line in each panel illustrates the one-to-one relative response to emissions reductions, whilst the dashed coloured lines are the linear regressions through each set of three model simulations, and they illustrate the actual responses to emissions reductions of a given precursor. The numbers show the corresponding relative responses to each emissions reduction (with respect to baseline).

In East Asia and North America, NH₃ concentrations increase similarly for either NO_x or SO_x reductions (Fig. 3), but in South Asia and Euro_Medi, NH₃ concentrations increase more with SO_x reductions than with NO_x reductions, which reflects the larger contribution of (NH₄)₂SO₄ than NH₄NO₃ to SIA in the latter two regions. However, the increase in NH₃ concentrations is relatively small compared to the extent of NO_x and SO_x emissions reductions: 40 % reductions in emissions of NO_x or SO_x only increase NH₃ concentrations in the four regions by 2 %–6 % or 6 %–9 %, respectively (Fig. 3, Table S2). The globally averaged increases in NH₃ concentrations for 40 % reductions in NO_x or SO_x emissions are 3 % and 9 %, respectively. Nevertheless, the NH₃ concentration decrease resulting from reductions in NH₃ emissions is offset by simultaneous effects of NO_x and SO_x emissions reductions when all three precursors are reduced together, as the sensitivities of regional average NH₃ concentrations to 40 % reductions in all three precursor emissions (38 %–39 % across the four regions) are smaller than their sensitivities to 40 % reductions in NH₃ emissions on its own (47 %–49 %). It is also noteworthy that the sensitivities of regional average NH₃ concentrations are essentially linear through 20 % and 40 % emissions reductions, irrespective of precursor, although the sensitivities are different between regions.

In contrast to NH₃, concentrations of NH${}_{\mathrm{4}}^{+}$ always decrease when an SIA precursor emission is reduced. Figure 2 shows that NH${}_{\mathrm{4}}^{+}$ concentrations in the most densely populated continents (e.g. eastern China, India, Europe, eastern America) respond strongly to emissions reductions in each SIA precursor, whilst they only respond to SO_x emissions reductions over oceans. This is related to the production of marine sulfate aerosol from dimethyl sulfide (DMS) and the lack of significant oceanic NO_x emissions sources, which means only (NH₄)₂SO₄ formation is important in marine SIA chemistry.

In addition, the impacts of NH₃ and SO_x emissions reductions on NH${}_{\mathrm{4}}^{+}$ concentrations over North Africa are significantly greater than from NO_x emissions reductions, indicating a dominance of (NH₄)₂SO₄ within SIA in this region. This is consistent with the results reported by Ge et al. (2022). They showed that large areas in North Africa are characterized by the SO${}_{\mathrm{4}}^{\mathrm{2}-}$-rich chemical domain for SIA formation, which means that NH₃ is predominantly taken up by SO${}_{\mathrm{4}}^{\mathrm{2}-}$, leaving no free NH₃ to react with HNO₃ to form NH₄NO₃. Given that emissions reductions in all three precursors individually lead to reductions in NH${}_{\mathrm{4}}^{+}$ concentrations, it is not surprising that the greatest simulated NH${}_{\mathrm{4}}^{+}$ reduction (5.87 µg m⁻³ (43 %) in East Asia) arises for the scenario with 40 % reductions in all three precursors collectively.

Figure 4 Changes in NO_x and fine NO${}_{\mathrm{3}}^{-}$ annual surface concentrations for 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

Figure 3 also shows that NH${}_{\mathrm{4}}^{+}$ sensitivities are essentially linear for emissions reductions up to 40 %, although responses again vary slightly with region. Among individual precursor reduction scenarios, regional average NH${}_{\mathrm{4}}^{+}$ concentrations in East Asia and Euro_Medi are most sensitive to SO_x emissions reductions and least sensitive to NO_x reductions, while NH${}_{\mathrm{4}}^{+}$ concentrations in North America are most sensitive to SO_x reductions and least sensitive to NH₃ reductions. In South Asia, NH${}_{\mathrm{4}}^{+}$ is characterized by strong sensitivity to SO_x emissions reductions but only relatively small sensitivities to NO_x and NH₃ emissions reductions. In the scenario of all three species reductions, all regions show relative sensitivities close to the one-to-one line. Another important observation from Fig. 3 is that reductions in NH${}_{\mathrm{4}}^{+}$ (4 %–18 %) in response to a 40 % NH₃ emissions reduction are much smaller than reductions of NH₃ concentrations (47 %–49 %) in these regions, which reflects the fact that these regions are so ammonia rich that reducing NH₃ emissions only has limited effects on NH${}_{\mathrm{4}}^{+}$ concentrations.

3.1.2 OXN

Figure 4 shows the spatial variations in the sensitivities of NO_x and fine nitrate annual mean surface concentrations to 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Regional average sensitivities in East Asia, South Asia, Euro_Medi, and North America are summarized in Fig. 5 and Table S3. Equivalent global maps and regional sensitivity plots for the responses of HNO₃ and coarse nitrate to the same emissions reductions are presented in Figs. S2 and S3 and Table S3.

In contrast to surface NH₃, whose concentrations are sensitive to reductions in emissions of each of NH₃, NO_x, and SO_x, surface NO_x concentrations only respond to NO_x emissions reductions and have negligible sensitivity to NH₃ and SO_x emissions reductions. The 20 % and 40 % reductions in NO_x emissions yield a global maximum of 25.6 µg m⁻³ (22 %) and 51.3 µg m⁻³ (44 %) reductions in surface NO_x concentrations over East Asia. The globally averaged reductions in NO_x concentrations for 20 % and 40 % reductions in NO_x emissions are 15 % and 30 %, respectively (same values for simultaneous reductions in all three precursors), whereas the sensitivities for NH₃ and SO_x emissions reduction are all 0 % (Table S3).

Figure 5 The absolute and relative sensitivities of regionally averaged annual mean surface concentrations of NO_x (upper row) and fine NO${}_{\mathrm{3}}^{-}$ (lower row) to 20 % and 40 % emissions reductions in NH₃ (blue), NO_x (orange), and SO_x (green), individually and collectively (red), for the four regions defined in Fig. 1. The solid grey line in each panel illustrates the one-to-one relative response to emissions reductions, whilst the dashed coloured lines are the linear regressions through each set of three model simulations, and they illustrate the actual responses to emissions reductions of a given precursor. The numbers show the corresponding relative responses to each emissions reduction (with respect to baseline).

Regionally (Fig. 5), East Asia shows the largest decreases in NO_x concentrations (45 %) in response to 40 % NO_x emissions reductions, followed by Euro_Medi and North America (36 %–38 %), and South Asia (23 %). These regional differences in sensitivities to NO_x emissions are due to regional differences in oxidation chemistry climate. Figure S4 shows the changes in annual mean surface concentrations of O₃ for the eight emissions reduction scenarios. Concentrations of O₃ in eastern China, western and central Europe, and the north-eastern US increase as NO_x emissions reduce, while O₃ in the rest of the world decreases as NO_x emissions reduce. In East Asia, the increased oxidant levels enhance NO_x chemical removal and result in a greater than one-to-one relative decrease in NO_x concentrations with NO_x emissions reductions. Consequently, the decrease in fine nitrate in East Asia is offset by enhanced chemical production, which leads to a lower than one-to-one sensitivity (e.g. a 40 % reduction in NO_x emissions gives a 33 % decrease in fine NO${}_{\mathrm{3}}^{-}$). In contrast, decreased oxidant levels in South Asia decrease the oxidation of NO_x, which partially offsets the decrease in NO_x concentrations induced by emissions reductions, causing a more efficient reduction in fine NO${}_{\mathrm{3}}^{-}$ concentrations than in NO_x in this region. The variation in regional atmospheric oxidizing capacity also alters the SO${}_{\mathrm{4}}^{\mathrm{2}-}$ formation processes; discussions of this are presented in Sects. 3.1.3 and 4. The situation is more complex for Euro_Medi and North America, as these regions include both positive and negative changes in O₃ concentrations with NO_x emissions reductions (Fig. S4), and they are not as NO_x rich as East Asia and South Asia. The effects of changes in oxidant levels on NO_x and fine NO${}_{\mathrm{3}}^{-}$ concentrations are therefore more localized and less apparent in regional averages. Clappier et al. (2021) reported this effect to be most distinct in the Po basin (Italy), western Germany, and the Netherlands in Europe, whilst for the United States, Tsimpidi et al. (2008) showed it only becomes pronounced in the north-east, both of which are consistent with our results.

For secondary OXN species, the sensitivities of HNO₃ and fine and coarse NO${}_{\mathrm{3}}^{-}$ to individual reductions in emissions of NH₃, NO_x, and SO_x are closely associated with SIA formation chemistry. The principal observation from Fig. 4 is that concentrations of fine NO${}_{\mathrm{3}}^{-}$ in all four regions decrease with reduced NH₃ and NO_x emissions but increase with reduced SO_x emissions. This is because H₂SO₄ and HNO₃ compete in their reactions with NH₃, and (NH₄)₂SO₄ is preferably formed over NH₄NO₃. Reductions in NH₃ emissions cause the equilibrium between HNO₃ and NH₃ to shift away from NH₄NO₃ production and therefore free more HNO₃ molecules. As a result, more HNO₃ is available to produce coarse nitrate aerosol, leading to a decrease in fine NO${}_{\mathrm{3}}^{-}$ but an increase in coarse NO${}_{\mathrm{3}}^{-}$ concentrations (Fig. S2). Reductions in NO_x emissions decrease HNO₃ and fine and coarse NO${}_{\mathrm{3}}^{-}$ concentrations globally. Although the increased oxidant levels that arise in some regions following NO_x emissions reductions enhance the chemical formation of these secondary species (as discussed above), NO_x emissions reductions of 20 % and 40 % are substantial enough to mean that the lower availability of NO_x to form NO${}_{\mathrm{3}}^{-}$ dominates the impact on NO${}_{\mathrm{3}}^{-}$ concentrations compared with the enhancement in oxidizing capacity. Reduced SO_x emissions leave more NH₃ to equilibrate with HNO₃ to form NH₄NO₃, leading to an increase in fine NO${}_{\mathrm{3}}^{-}$ concentrations but a decrease in coarse NO${}_{\mathrm{3}}^{-}$ concentrations, as the former takes more HNO₃. It is notable that the increase in fine NO${}_{\mathrm{3}}^{-}$ concentrations is relatively small compared to the extent of SO_x emissions reductions. For example, the maximum increase in fine NO${}_{\mathrm{3}}^{-}$ resulting from 40 % reductions in SO_x emissions is 1.71 µg m⁻³ (16 %), in East Asia. The regional average increases in fine NO${}_{\mathrm{3}}^{-}$ concentrations for 40 % SO_x emissions reductions are 8 % in East Asia, South Asia, and Euro_Medi and 4 % in North America.

The differences in regional average sensitivities of HNO₃ and fine and coarse NO${}_{\mathrm{3}}^{-}$ are highlighted more clearly in Figs. 5 and S2. Fine NO${}_{\mathrm{3}}^{-}$ in East Asia is equally sensitive to NO_x and NH₃ emissions reductions (33 % and 32 % decreases for 40 % NO_x and NH₃ emissions reductions, respectively), while it is more sensitive to NO_x emissions reductions than to NH₃ emissions reductions in South Asia (45 % and 39 %), Euro_Medi (41 % and 33 %), and North America (42 % and 26 %). In terms of absolute concentration changes, the reductions in fine NO${}_{\mathrm{3}}^{-}$ over East Asia in response to 40 % NH₃ and NO_x emissions reductions (1.62–1.65 µg m⁻³) are more than 3 times larger than reductions in other regions (0.23–0.47 µg m⁻³). On the other hand, if NH₃ emissions are reduced, then the increases in HNO₃ and coarse NO${}_{\mathrm{3}}^{-}$ concentrations in East Asia (15 % increases for 40 % NH₃ reductions) are much larger than the increases in the other three regions (2 %–6 %). All these differences between East Asia and the other three regions reflect the larger abundance of NH₄NO₃ in SIA over East Asia. This is demonstrated in Fig. S8, which shows that the contribution of fine NO${}_{\mathrm{3}}^{-}$ to PM_2.5 in the baseline is greatest in East Asia (19 %, 5.21 µg m⁻³), followed by Euro_Medi (12 %, 1.22 µg m⁻³), North America (11 %, 0.86 µg m⁻³), and South Asia (3 %, 0.93 µg m⁻³). Detailed discussion on regional SIA composition is presented in Sect. 3.2.

3.1.3 OXS

Figure 6 shows the global variation in the sensitivities of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ annual mean surface concentrations to 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Figure 7 and Table S4 summarize the sensitivities of the regionally averaged SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations to the emissions reductions for the four regional domains. Concentrations of SO₂ increase in response to reduced NH₃ emissions particularly over East Asia, South Asia, Europe, and North America. The largest increases in SO₂ resulting from 20 % and 40 % NH₃ emissions reductions are, respectively, in South East Asia (3.62 µg m⁻³, 9 %) and East Asia (8.11 µg m⁻³, 20 %) (Fig. 6). The response of secondary SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations to NH₃ emissions reductions varies substantially across the world. In north-eastern China and Europe, SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations increase, whilst in southern China, India, and United States they decrease. However, the magnitudes of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentration changes are much smaller than for SO₂. The maximum increases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (located in southern China) are only 0.37 µg m⁻³ (2 %) and 0.89 µg m⁻³ (5 %) for 20 % and 40 % NH₃ emissions reductions, respectively. The maximum decreases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations, in north-eastern China, are 0.62 µg m⁻³ (6 %) and 1.30 µg m⁻³ (12 %), respectively. Regionally averaged, East Asia exhibits the largest increase in SO₂ for 40 % NH₃ emissions reductions (16 %) (Fig. 7), followed by North America (14 %), South Asia (14 %), and Euro_Medi (10 %), whereas increases in regional average SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations are only in the range of 0 %–2 %.

Figure 6 Changes in SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ annual surface concentrations for 20 % and 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

Figure 7 The absolute and relative sensitivities of regionally averaged annual mean surface concentrations of SO₂ (upper row) and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (lower row) to 20 % and 40 % emissions reductions in NH₃ (blue), NO_x (orange), and SO_x (green), individually and collectively (red), for the four regions defined in Fig. 1. The solid grey line in each panel illustrates the one-to-one relative response to emissions reductions, whilst the dashed coloured lines are the linear regressions through each set of three model simulations, and they illustrate the actual responses to emissions reductions of a given precursor. The numbers show the corresponding relative responses to each emissions reduction (with respect to baseline).

Figures S5 and 13 show the global variation in the sensitivities to precursor emissions reductions of total deposition of SO₂ and of the wet and dry deposition of all OXS components, respectively. When NH₃ emissions are reduced, SO₂ total deposition decreases over all of East Asia, South Asia, Europe, and North America (Fig. S5); Fig. 13 shows that the decrease is driven by reduced SO₂ dry deposition. Several studies have shown that the non-stomatal canopy uptake resistance of SO₂ (the inverse of the SO₂ dry deposition velocity) is positively correlated to the molar “acidity ratio” $a_{\mathrm{SN}}=\frac{\left[{\mathrm{SO}}_{\mathrm{2}}\right]}{\left[{\mathrm{NH}}_{\mathrm{3}}\right]}$ (Smith et al., 2000; Erisman et al., 2001; Fowler et al., 2009; Massad et al., 2010), a process that is included in the EMEP MSC-W model (Simpson et al., 2012). Reduced NH₃ concentrations therefore increase the acidity ratio and hence decrease the rate of SO₂ dry deposition and increase the SO₂ surface concentrations in those regions where this effect is significant. The SO${}_{\mathrm{4}}^{\mathrm{2}-}$ responses to NH₃ emissions reductions are related to changes in atmospheric acidity as well. The aqueous-phase oxidation of SO₂ by O₃, which is one of the major pathways for SO${}_{\mathrm{4}}^{\mathrm{2}-}$ production, is significantly pH dependent. In general, the oxidation rate decreases with decreased pH (Penkett et al., 1979; Maahs, 1983; Liang and Jacobson, 1999; Hattori et al., 2021), a process that is incorporated into the EMEP MSC-W model (Simpson et al., 2012). As pointed out by Ge et al. (2022), Europe and north-eastern China are much less ammonia rich than India, which means that 20 % and 40 % NH₃ emissions reductions are substantial enough to decrease the pH in the former two regions, leading to decreases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ production. However, given that SO₂ levels increase and that there are still other effective oxidation pathways (e.g. OH, H₂O₂) which are independent of pH (McArdle and Hoffmann, 1983; Hoffmann, 1986; Seinfeld and Pandis, 2016), the decreases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in Europe and north-eastern China due to NH₃ emissions reductions are very small anyway. In contrast, since India is so ammonia rich, even 40 % reductions in NH₃ emissions do not significantly alter the pH in this region. As a result, SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in India increase slightly due to the higher availability of its precursor SO₂.

The impacts of NO_x emissions reductions on SO₂ concentrations show inverse trends in different regions (Fig. 7), which reflects regional differences in atmospheric oxidation chemistry. The decreased SO₂ concentrations (maximum reduction: 1.72 µg m⁻³, 6 %) in eastern China, Europe, and the north-eastern United States (Fig. 6) can be explained by the enhanced O₃ concentrations in these regions arising from the reduced NO + O₃ reaction in these high NO_x regions (Fig. S4). As a result, SO₂ is more readily oxidized to SO${}_{\mathrm{4}}^{\mathrm{2}-}$, leading to increased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (maximum increase for 40 % NO_x emissions reductions: 1.59 µg m⁻³ (10 %)) in these regions. This positive response of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ to NO_x emissions reductions is also reported in regional studies (Botha et al., 1994; Li et al., 2006; Sheng et al., 2018; Fang et al., 2019; W. Ge et al., 2021). In contrast, India, north-eastern Africa, and southern Africa show increased SO₂ (maximum increase: 0.29 µg m⁻³, 1 %) but decreased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (maximum decrease: 0.73 µg m⁻³, 9 %) as NO_x emissions reduce, which can be explained by the parallel decrease in O₃ concentrations in these regions (Fig. S4). However, these concentration changes are very localized and, from a regional average perspective, are relatively small compared to the extent of emissions reductions applied. For example, in East Asia, the region with the largest response, there is only a 5 % decrease in regional average SO₂ concentration (Fig. 7), and 3 % increase in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentration, for a 40 % reduction in NO_x emissions. For other regions, the SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ regional average concentration changes are even smaller (from −4 % to 2 %). The global average sensitivities of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ annual mean concentrations to 20 % and 40 % NO_x emissions reductions are only in the range of 0 %–2 % (Table S4).

Under reductions of SO_x emissions (and of all three precursors together), both SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ show almost one-to-one reductions, indicating that SO_x emissions reductions are crucial for reducing both primary and secondary OXS pollutants and, in the case of reductions of all three precursors simultaneously, readily sufficient to dominate over any tendency for NH₃ and NO_x emissions reductions to increase OXS species concentrations. A 40 % reduction in SO_x emissions leads to a maximum SO₂ decrease of 39.8 µg m⁻³ (40 %) in northern Russia and a maximum SO${}_{\mathrm{4}}^{\mathrm{2}-}$ decrease of 7.70 µg m⁻³ (39 %) in south-eastern China (Fig. 6). For the four regions, average SO₂ concentrations decrease by 22 %–24 % and 42 %–45 %, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations decrease by 17 %–19 % and 34 %–38 %, in response to 20 % and 40 % SO_x emissions reductions, respectively (Fig. 7). For the 20 % and 40 % reductions in all three precursors together, regionally averaged SO₂ decreases by 19 %–20 % and 38 %–41 %, respectively, and regionally averaged SO${}_{\mathrm{4}}^{\mathrm{2}-}$ decreases by 17 %–20 % and 35 %–40 %, respectively.

3.2 Sensitivity of PM_2.5 concentrations

Figure 8 shows the global distribution of the dominant sensitivity of PM_2.5 towards a 40 % reduction in NH₃, NO_x, or SO_x emissions individually. An annual mean PM_2.5 concentration threshold of 5 µg m⁻³ has been applied in order to focus attention on more polluted areas where PM_2.5 concentrations in the baseline simulation are above the latest WHO PM_2.5 air quality guideline (AQG) (WHO, 2021). Section S2 and Fig. S7 in the Supplement present seasonal variations in global PM_2.5 sensitivity regimes.

Figure 8 Spatial variation in sensitivity regime of PM_2.5 mitigation based on data from 40 % individual reductions in emissions of NH₃, NO_x, or SO_x. The regime is defined according to the precursor that yields the greatest decreases in grid PM_2.5 concentration: NH₃ sensitive (yellow), NO_x sensitive (blue), and SO_x sensitive (green). Model grids with baseline annual mean PM_2.5 concentrations < 5 µg m⁻³ are masked out.

The principal observation from Fig. 8 is that the sensitivity of PM_2.5 to reductions in emissions of individual precursors is highly geographically variable. SO_x-sensitive regimes are found in South East Asia, South Asia, Africa, and Central America. NO_x-sensitive regimes are observed in south-eastern China, France, Germany, most eastern European countries, the central and eastern United States, and northern and central parts of South America. Only a few small regions are NH₃ sensitive: these include eastern coastal areas around China, the UK and its surrounding seas, southern Scandinavia, and western Russia. The difference in PM_2.5 sensitivity between northern Europe and the rest of Europe demonstrates that NH₃ has become the limiting factor for SIA formation in northern Europe. This greater leverage of NH₃ emissions on PM_2.5 mitigation in this region is due to the effective emissions controls on all SIA precursor emissions here (see also Sect. 4 discussion) (Tørseth et al., 2012; AQEG, 2015; Vieno et al., 2016; Ciarelli et al., 2019; Theobald et al., 2019). In contrast, South Asia is so ammonia rich that reducing NH₃ concentrations has little impact on PM_2.5 (Ge et al., 2022). The situation in northern Europe exemplifies what to expect in terms of future policy making for the rest of the world. Furthermore, many marine areas are characterized as SO_x sensitive but for a different reason than the SO_x-sensitive regime in South Asia. In the marine areas, fine nitrate and ammonium aerosols are relatively small compared to sulfate aerosols; therefore, reductions in NO_x and NH₃ emissions hardly affect SIA formation. In fact, sulfate aerosol derived from oceanic emissions of DMS rather than from anthropogenic emissions is the major contributor to marine PM_2.5 (Quinn and Bates, 2011; Hoffmann et al., 2016; Novak et al., 2022).

Figure 9 Changes in PM_2.5 annual mean surface concentrations for 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. The blue star in each map locates the minimum difference within each region.

Detailed examination of the magnitudes of the PM_2.5 sensitivities in each location to each precursor (rather than just their ranking) reveals more complicated regional characteristics. Figure 9 shows the spatial variabilities in PM_2.5 sensitivities to 40 % reductions in emissions of NH₃, NO_x, and SO_x, individually and collectively, for the four world regions. The regionally averaged PM_2.5 sensitivities are summarized in Fig. 10 and Table S5. PM_2.5 concentrations in East Asia show comparable sensitivities to individual emissions reductions in NH₃, NO_x, and SO_x, with the impacts of NH₃ and NO_x emissions reductions being more concentrated in continental areas than those for SO_x emissions reductions (Fig. 9). A 40 % reduction in NO_x emissions yields a maximum decrease in PM_2.5 of 11.5 µg m⁻³ (12 %) over southern China, while 40 % reductions in NH₃ and SO_x yield slightly smaller maximum PM_2.5 decreases of 8.51 µg m⁻³ (10 %) and 8.78 µg m⁻³ (9 %), respectively. The regional average sensitivities of PM_2.5 concentrations in East Asia to 40 % reductions in individual precursors are ∼ 8 % (Fig. 10).

Figure 10 The absolute and relative sensitivities of regionally averaged annual mean surface concentrations of PM_2.5 to 20 % and 40 % emissions reductions in NH₃ (blue), NO_x (orange), and SO_x (green), individually and collectively (red), for the four regions defined in Fig. 1. The solid grey line in each panel illustrates the one-to-one relative response to emissions reductions, whilst the dashed coloured lines are the linear regressions through each set of three model simulations, and they illustrate the actual responses to emissions reductions of a given precursor. The numbers show the corresponding relative responses to each emissions reduction (with respect to baseline).

In contrast, PM_2.5 concentrations in South Asia only show significant responses to SO_x emissions reductions, whilst NH₃ emissions reductions have little effect, which is consistent with our previous finding that South Asia has the most ammonia-rich chemical climate for SIA formation (Ge et al., 2022). The dominant proportion of (NH₄)₂SO₄ in SIA compared to NH₄NO₃ in South Asia (Fig. S8) also explains the small sensitivity of PM_2.5 in this region to NO_x emissions reductions. The maximum PM_2.5 decrease (7.02 µg m⁻³, 9 %) for 40 % SO_x emissions reductions is more than 3 times larger than the maximum PM_2.5 decrease (2.20 µg m⁻³, 4 %) for 40 % NH₃ emissions reductions. The decreases in regionally averaged PM_2.5 concentrations in South Asia in response to 40 % reductions in emissions of individual precursors are in the order of 10 % for SO_x, 3 % for NO_x, and 1 % for NH₃.

In the Euro_Medi region, PM_2.5 sensitivities vary from north to south. Northern and central Europe are most sensitive to NH₃ and NO_x emissions reductions, for which the maximum decrease in PM_2.5 is ∼ 2.6 µg m⁻³ (16 %) for 40 % reductions, while the Mediterranean is more sensitive to SO_x emissions reductions, for which the maximum decrease in PM_2.5 is 2.98 µg m⁻³ (12 %) for 40 % reductions. Regionally averaged, however, the PM_2.5 concentrations in Euro_Medi show comparable sensitivities to the three precursors, with decreases in the range 5 %–8 % for 40 % emissions reductions in individual precursors.

Over North America, the eastern US shows larger sensitivities of PM_2.5 to all emissions reduction scenarios than the western US, and reductions in NO_x emissions yield larger decreases in PM_2.5 than reductions in SO_x and NH₃ emissions. The maximum decrease in PM_2.5 derived from 40 % reductions in NO_x emissions is 3.10 µg m⁻³ (16 %); for 40 % reductions in NH₃ and SO_x emissions the maximum PM_2.5 decreases are 2.37 µg m⁻³ (14 %) and 1.35 µg m⁻³ (10 %), respectively. The regional average sensitivities of PM_2.5 concentrations in North America to 40 % reductions in emissions of individual precursors decreases in a slightly different order: NO_x (8 %), SO_x (7 %), and NH₃ (4 %).

Figure 10 shows that 20 % emissions reductions in any precursor lead to decreases in regionally averaged PM_2.5 concentrations, although the PM_2.5 sensitivities vary substantially with precursor and region. Given the non-one-to-one chemical responses of SIA components to reductions in emissions in individual precursors discussed in Sect. 3.1, even 20 % reductions appear substantial enough to ensure that decreased SIA formation due to decreased precursor emissions dominates over any disbenefits to SIA formation from, for example, increases in oxidant levels induced by NO_x emissions reductions. For instance, 20 % reductions in NO_x emissions still cause a decrease of 0.77 µg m⁻³ (3 %) in regional average PM_2.5 in East Asia, despite increasing regional average SO${}_{\mathrm{4}}^{\mathrm{2}-}$ by 0.10 µg m⁻³ (2 %), because it decreases regional average NH${}_{\mathrm{4}}^{+}$ and fine NO${}_{\mathrm{3}}^{-}$ by greater amounts (0.17 µg m⁻³ (5 %) and 0.72 µg m⁻³ (14 %), respectively). Similarly, 20 % reductions in SO_x emissions decrease regional average PM_2.5 in East Asia by 1.15 µg m⁻³ (4 %) because the decreases in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ (1.06 µg m⁻³, 19 %) and NH${}_{\mathrm{4}}^{+}$ (0.32 µg m⁻³, 9 %) caused by reduced (NH₄)₂SO₄ formation are larger than the increase in fine NO${}_{\mathrm{3}}^{-}$ (0.20 µg m⁻³, 4 %) due to elevated NH₄NO₃ formation. On the other hand, the mitigation of PM_2.5 by reducing emissions of all three precursors together is impacted by these non-one-to-one chemical responses as well, which causes the net decrease in regional average PM_2.5 derived from reductions in all three precursors to be smaller than the sum of individual PM_2.5 decreases derived from reductions in emissions of precursors individually. For example, 40 % reductions in NH₃, NO_x, and SO_x emissions individually decrease regional average PM_2.5 in East Asia by 2.03 µg m⁻³ (7 %), 1.89 µg m⁻³ (7 %), and 2.33 µg m⁻³ (8 %), respectively (sum of the three: 6.25 µg m⁻³), while the decrease in regional average PM_2.5 derived from 40 % reduction in all three precursors simultaneously is 5.59 µg m⁻³ (20 %).

The 40 % reduction in emissions of all three species yields a maximum decrease in PM_2.5 of 23.9 µg m⁻³ (25 %) over East Asia, followed by 10.4 µg m⁻³ (17 %) in South Asia, 5.57 µg m⁻³ (22 %) in Euro_Medi, and 5.05 µg m⁻³ (28 %) in North America. The regional average sensitivity of PM_2.5 concentrations to 20 % and 40 % reductions in emissions of all three species decreases in the order of East Asia (10 % and 20 % for 20 % and 40 % reductions, respectively), Euro_Medi (9 % and 17 %), North America (8 % and 17 %), and South Asia (7 % and 13 %). This trend is related to differences in the contribution of SIA to PM_2.5 in the different regions. Figure S8 shows the mass contributions of individual PM_2.5 components to the regional average concentration of PM_2.5 in the baseline and the 40 % emissions reductions scenarios. SIA components in the baseline account for over half of PM_2.5 in East Asia (52 %), followed by Euro_Medi (42 %), North America (35 %), and South Asia (31 %), which explains why reductions in emissions of all three SIA precursors are most efficient for the mitigation of PM_2.5 in East Asia but least efficient in South Asia. In fact, primary PM_2.5 is the largest contributor to PM_2.5 in South Asia, so reducing these emissions will be the most efficient way of abating PM_2.5 pollution in this region. It is noteworthy that in the scenario of 40 % reductions in all three species, SIA is still the largest contributor to PM_2.5 in East Asia, while primary PM_2.5 and Rest (mainly secondary organic aerosol) become the dominant contributors in other regions. Even with 40 % reductions in all three SIA precursors, none of the four regions has its regional average PM_2.5 concentration decreased to below 5 µg m⁻³. Euro_Medi (8.4 µg m⁻³ after 40 % reductions) and North America (6.5 µg m⁻³) are the closest, whilst East Asia (21.8 µg m⁻³) and South Asia (27.0 µg m⁻³) are still far away from achieving the latest WHO AQG for PM_2.5. Therefore, reductions in emissions of primary PM_2.5 and in VOCs are also required to achieve further PM_2.5 reductions in all regions or even greater reductions in SIA precursors than simulated here.

3.3 Sensitivity of N_r and S_r deposition

The impacts of reductions in emissions of NH₃, NO_x, and SO_x on total amounts of N and S deposition are straightforward because these must match the emissions mass changes in N and S. However, the relative amounts and spatial pattern of the individual components of N and S deposition are impacted.

Figure 11 Changes in wet (WDEP) and dry deposition (DDEP) of NH₃ and NH${}_{\mathrm{4}}^{+}$ for 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

Figure 11 shows the spatial variations in the sensitivities of the wet and dry deposition of the RDN components NH₃ and NH${}_{\mathrm{4}}^{+}$ to 40 % reductions in global emissions of NH₃, NO_x, and SO_x, individually and collectively. Figure 12 shows similar for deposition of the OXN components NO_x, HNO₃, TNO${}_{\mathrm{3}}^{-}$ (total NO${}_{\mathrm{3}}^{-}$, the sum of fine and coarse NO${}_{\mathrm{3}}^{-}$), and Rest (the sum of other oxidized species). Both wet and dry deposition of NH₃ show negative responses to emissions reductions in NH₃ and all three precursors together but positive responses to emissions reductions in NO_x and SO_x, which is consistent with the responses of surface NH₃ concentrations to these emissions reductions (Sect. 3.1.1). This is because reduced NO_x and SO_x emissions lead to decreased concentrations of acidic species in the atmosphere, resulting in more NH₃ remaining in the gas phase and greater NH₃ deposition over continents. In contrast, global NH${}_{\mathrm{4}}^{+}$ wet and dry deposition decreases in all emissions reduction scenarios, which is also in line with the decreased NH${}_{\mathrm{4}}^{+}$ concentrations in all scenarios.

The changes in deposition of the OXN components to emissions reductions are more complicated (Fig. 12). The two species with the largest variation in deposition across the emissions reduction scenarios are HNO₃ and TNO${}_{\mathrm{3}}^{-}$, which is due to their large contributions to total OXN deposition in most world regions (Ge et al., 2022). For reductions in emissions of NO_x and all three species, all OXN deposition components show clear decreasing trends due to the strong reduction in their precursor emissions.

Figure 12 Changes in wet (WDEP) and dry deposition (DDEP) of NO_x, HNO₃, TNO${}_{\mathrm{3}}^{-}$ (fine + coarse NO${}_{\mathrm{3}}^{-}$), and Rest OXN species for 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

In response to NH₃ emissions reductions, Fig. 12 shows that HNO₃ wet and dry deposition increases in eastern China, northern India, Europe, and eastern North America, whereas the wet and dry deposition of TNO${}_{\mathrm{3}}^{-}$ decreases in these regions. Further examination of fine and coarse NO${}_{\mathrm{3}}^{-}$ deposition differences in Fig. S9 shows that the decrease in TNO${}_{\mathrm{3}}^{-}$ deposition is driven by the decrease in fine NO${}_{\mathrm{3}}^{-}$ wet and dry deposition, while coarse NO${}_{\mathrm{3}}^{-}$ wet and dry deposition in the four regions actually increases. The reduction in NH₃ emissions decreases NH₄NO₃ formation and therefore liberates more HNO₃; as a result, more OXN deposits in the form of HNO₃ rather than NO${}_{\mathrm{3}}^{-}$ in these regions. In contrast, impacts of SO_x emissions reductions on HNO₃ and TNO${}_{\mathrm{3}}^{-}$ deposition are the opposite of impacts of NH₃ emissions reductions. As discussed in Sect. 3.1.2, decreased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations promote the formation of NH₄NO₃, which results in decreased wet and dry deposition of HNO₃ and consequently increased TNO${}_{\mathrm{3}}^{-}$ deposition (driven by increased fine NO${}_{\mathrm{3}}^{-}$ deposition, Fig. S9) in eastern China, northern India, Europe, and eastern North America.

Compared to HNO₃ and TNO${}_{\mathrm{3}}^{-}$, the responses of NO_x and Rest OXN deposition to reductions in NH₃ and SO_x emissions are considerably smaller. For instance, the maximum increases in HNO₃ wet and dry deposition in response to 40 % NH₃ emissions reductions are 328 mgN m⁻² (85 %) and 248 mgN m⁻² (48 %), respectively, whereas the maximum increases in Rest OXN wet and dry deposition are only 1.35 mgN m⁻² (4 %) and 1.64 mgN m⁻² (4 %), respectively. Also, in the 40 % NH₃ emissions reduction scenario, the maximum decreases in TNO${}_{\mathrm{3}}^{-}$ wet and dry deposition are 359 mgN m⁻² (25 %) and 175 mgN m⁻² (28 %), respectively, whereas the maximum decreases in NO_x dry deposition are only 4.22 mgN m⁻² (1 %).

Figure 13 Changes in wet (WDEP) and dry deposition (DDEP) of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ for 40 % emissions reductions in NH₃, NO_x, and SO_x, individually and collectively. Red and green dots in each map locate the minimum and maximum difference, respectively.

The spatial variations in the sensitivities of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ wet and dry deposition to the precursor emissions reductions are shown in Fig. 13. Increased SO₂ wet deposition is observed globally in response to NH₃ emissions reductions, whilst SO₂ dry deposition decreases over the continents but increases over the oceans. The non-stomatal canopy resistance of SO₂ is positively correlated to the molar acidity ratio a_SN (Smith et al., 2000; Erisman et al., 2001; Fowler et al., 2009; Massad et al., 2010; Simpson et al., 2012). Reductions in NH₃ emissions increase the a_SN, which therefore increases the canopy resistance of SO₂ and decreases SO₂ dry deposition over the continents. The 40 % reduction in NH₃ emissions yields a global maximum increase in SO₂ wet deposition of 131 mgS m⁻² (44 %) and a maximum decrease in SO₂ dry deposition of 600 mgS m⁻² (24 %). The overall effect of increased SO₂ wet deposition and decreased SO₂ dry deposition is a decreased SO₂ total deposition over populated continents (where NH₃ emissions are high) and increased SO₂ total deposition over oceans (Fig. S5). The maximum decrease of SO₂ total deposition located in southern China is 586 mgS m⁻² (18 %) for 40 % NH₃ reduction.

The sensitivity of total deposition of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ to NH₃ emissions reductions (Fig. S6) follows the trend in the sensitivity of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (Sect. 3.1.3). The responses of wet and dry deposition of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ to NH₃ emissions reductions are similar (Fig. 13). In general, decreased wet and dry SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition appears over Europe, north-eastern China, and the north-eastern US, while increased wet and dry deposition occurs in the rest of the world. As also for their concentration sensitivities, the magnitudes of the SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition responses are much smaller than for SO₂. For 40 % NH₃ emissions reduction, the maximum decrease of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ total deposition is 42 mgS m⁻² (9 %), which is an order of magnitude smaller than that of SO₂ total deposition.

For NO_x emissions reductions, both wet and dry deposition of SO₂ generally show decreases in eastern China, Europe, and the north-eastern US but increases in the rest of the world, which is contrary to the responses of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ wet and dry deposition. This is related to enhanced chemical conversion of SO₂ to SO${}_{\mathrm{4}}^{\mathrm{2}-}$ due to increased atmospheric oxidizing capacity over eastern China, Europe, and the north-eastern US (details in Sect. 3.1.3). The maximum decrease in SO₂ wet and dry deposition in response to 40 % reductions in NO_x emissions is ∼ 65 mgS m⁻², while the maximum increase in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ wet and dry deposition is ∼ 50 mgS m⁻².

For reductions in SO_x and in all three precursors collectively, decreased wet and dry deposition of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ is observed globally, and the 40 % reductions in these two scenarios yield similar global maximum decreases in SO₂ deposition (wet: ∼ 937 mgS m⁻², 41 %; dry: ∼ 1338 mgS m⁻², 43 %) and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition (wet: ∼ 828 mgS m⁻², 38 %; dry: ∼ 150 mgS m⁻², 39 %).

Figure 14 The absolute sensitivities (emission reduction – baseline) of regional total deposition (wet and dry) of NH₃, NH${}_{\mathrm{4}}^{+}$, NO_x, HNO₃, TNO${}_{\mathrm{3}}^{-}$ (fine and coarse NO${}_{\mathrm{3}}^{-}$), SO₂, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ to 40 % emissions reductions in NH₃ (top row), NO_x (second row), and SO_x (third row), individually and collectively (bottom row), for the four regions defined in Fig. 1. The left-hand y axis in each panel is for RDN and OXN species, while the right-hand y axis is for OXS species.

Figure 14 shows the differences of regional total deposition of individual species between baseline and the 40 % emissions reduction scenarios. The regional total RDN, OXN, and OXS deposition and the contributions of individual components are listed in Tables S6 and S7. As expected, the responses of regional total RDN, OXN, and OXS deposition are essentially linear through 20 % and 40 % reductions in their corresponding precursor and largely insensitive to reductions in the other precursors, although there are slight differences between regions. For example, 20 % and 40 % reductions in NH₃ emissions, respectively, yield 20 %–21 % and 39 %–41 % decreases in total deposition of RDN in the four regions, whilst having no effect on regional total OXN and OXS deposition. The contributions of different deposition components do, however, vary with emission changes, and the different lifetimes of the deposition components contribute to the small variabilities in responses of total RDN, OXN, and OXS deposition to emissions reductions via differences in the transport through the regional boundaries.

A 40 % reduction in NH₃ emissions produces a decrease of 3.58 TgN yr⁻¹ (51 % of regional total deposition of the same species in baseline; similarly, hereinafter) in NH₃ total deposition and 1.16 TgN yr⁻¹ (24 %) in NH${}_{\mathrm{4}}^{+}$ total deposition over East Asia (Fig. 14), which causes the contribution of NH₃ to RDN deposition to decrease from 59 % (baseline) to 47 % (after 40 % NH₃ reduction). In other regions, reductions in NH₃ emissions also decrease NH₃ deposition more effectively than NH${}_{\mathrm{4}}^{+}$ deposition. Also in East Asia, a 40 % NH₃ emissions reduction increases HNO₃ deposition by 0.61 TgN yr⁻¹ (20 %) and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition by 0.06 TgS yr⁻¹ (2 %) but decreases TNO${}_{\mathrm{3}}^{-}$ deposition by 0.55 TgN yr⁻¹ (14 %) and SO₂ deposition by 0.12 TgS yr⁻¹ (3 %). As a result, the contribution of HNO₃ deposition to total OXN deposition increases from 30 % to 37 %, corresponding to a decrease in TNO${}_{\mathrm{3}}^{-}$ contribution from 54 % to 47 %, whereas changes in contributions of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ to OXS deposition are very small (Table S7). In other world regions, such changes in total deposition of HNO₃, TNO${}_{\mathrm{3}}^{-}$, SO₂, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ derived from NH₃ emissions reductions are similar but of a smaller magnitude.

For 40 % NO_x emissions reductions, TNO${}_{\mathrm{3}}^{-}$ deposition shows the largest decrease in East Asia (1.57 TgN yr⁻¹, 35 %), South Asia (0.44 TgN yr⁻¹, 26 %), and Euro_Medi (0.71 TgN yr⁻¹, 37 %), whilst HNO₃ deposition shows the largest decrease in North America (0.56 TgN yr⁻¹, 40 %). In contrast, the sensitivities of NO_x dry deposition to NO_x emissions reductions are very small. The contributions of individual OXN deposition components remain fairly constant in all regions. Furthermore, East Asia, Euro_Medi, and North America show a 5 % increase in the contribution of NH₃ deposition to total RDN deposition and a corresponding 5 % decrease in NH${}_{\mathrm{4}}^{+}$ contribution for 40 % NO_x emissions reductions, which reflects a small shifting of gas–aerosol partitioning for RDN as well. This kind of change in RDN deposition contribution is 2 % for South Asia. The impacts of NO_x emissions reductions on OXS deposition compositions are very small.

The 40 % reductions in SO_x emissions yield 3.24 TgS yr⁻¹ (39 %), 1.07 TgS yr⁻¹ (38 %), 1.17 TgS yr⁻¹ (33 %), and 0.78 (37 %) TgS yr⁻¹ decreases in OXS deposition over East Asia, South Asia, Euro_Medi, and North America, respectively. SO_x emissions reductions cause larger decreases in SO₂ deposition than in SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition in East Asia and Euro_Medi, while SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ deposition in South Asia and North America show similar sensitivities. This is associated with slightly greater proportions of SO₂ (56 %–58 %) to OXS deposition in the former regions than in the latter regions, and these proportions are not affected by SO_x emissions reductions (Table S7). NH₃ and NH${}_{\mathrm{4}}^{+}$ deposition is moderately sensitive to SO_x emissions reductions in the four regions. An increase of 0.75 TgN yr⁻¹ (11 %) in NH₃ total deposition and a decrease of 0.71 TgN yr⁻¹ (14 %) in NH${}_{\mathrm{4}}^{+}$ total deposition for 40 % reductions in SO_x emissions is observed over East Asia. For South Asia, Euro_Medi, and North America, the increases in NH₃ total deposition due to 40 % SO_x reductions are 0.51 TgN yr⁻¹ (12 %), 0.25 TgN yr⁻¹ (11 %), and 0.23 TgN yr⁻¹ (12 %), respectively. This is because reduced SO_x emissions lead to reductions in (NH₄)₂SO₄ formation, which then cause increased NH₃ but decreased NH${}_{\mathrm{4}}^{+}$ concentrations. Another side effect of SO_x emissions reductions in East Asia is a slight decrease in HNO₃ deposition (0.25 TgN yr⁻¹, 10 %) and an equivalent increase in TNO${}_{\mathrm{3}}^{-}$ deposition (0.21 TgN yr⁻¹, 5 %). The equivalent deposition changes are considerably smaller in the other three regions, which again indicates a larger amount of NH₄NO₃ pollution in East Asia than other regions.

For collective reductions in emissions of all precursors, the changes in deposition of each species reflect net effects of individual reductions in emissions of NH₃, NO_x, and SO_x. For instance, Fig. 14 shows that the decrease in NH₃ deposition in East Asia derived from 40 % reduction in emissions of all three species (2.58 TgN yr⁻¹, 36 %) is smaller than that from individual NH₃ emissions reductions (3.58 TgN yr⁻¹, 51 %) due to the compensating effects of simultaneous NO_x and SO_x emissions reductions in the former scenario. In contrast, the decrease in NH${}_{\mathrm{4}}^{+}$ deposition in East Asia for 40 % emissions reductions in all three species (2.11 TgN yr⁻¹, 43 %) is almost double that from individual NH₃ emissions reduction scenarios (1.16 TgN yr⁻¹, 24 %). The variations in chemical forms of RDN, OXN, and OXS deposition affect where they deposit as well, since N_r and S_r species have different lifetimes, and a shorter lifetime causes a more localized deposition. Many studies show that NH₃ and HNO₃ have shorter lifetimes than NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$ (Xu and Penner, 2012; Hauglustaine et al., 2014; Bian et al., 2017; Ge et al., 2022). The abatement of total N (TN = RDN + OXN) and S deposition within a certain region is partially offset by this more localized deposition pattern, especially in South Asia and Euro_Medi. The 40 % reductions in all three species emissions yield a 34 % (2.93 TgN yr⁻¹) decrease in regional TN deposition in South Asia and a 34 % (1.18 TgS yr⁻¹) decrease in regional OXS deposition in Euro_Medi, which means that less N_r and S_r pollution is transported outside these regions and more is deposited locally.

4 Discussions

4.1 Model uncertainty

This study uses global model simulations for eight emissions reduction scenarios to investigate the geographical variation in the sensitivities of N_r, S_r, and PM_2.5 to emissions reductions in inorganic precursor gases. Although the EMEP MSC-W ACTM is state of the art and widely used in scientific research and policy development, the analyses presented in this study are based on data from a single modelling system, and their accuracy is therefore subject to any uncertainties in the EMEP MSC-W ACTM's specific parameterization of atmospheric processes, as well as uncertainties in the chosen emission inventory and meteorology input. However, the evaluation of surface concentrations and wet deposition of N_r and S_r species from this model configuration for the same year against global measurements from 10 monitoring networks (Y. Ge et al., 2021) has demonstrated the model's capability to capture the overall spatial variations in annual concentrations of NH₃, NH${}_{\mathrm{4}}^{+}$, NO₂, HNO₃, fine NO${}_{\mathrm{3}}^{-}$, SO₂, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and their wet deposition in East Asia, South East Asia, Europe, and North America. No nationally compiled data for these species are publicly available in Oceania, South Asia, Africa, and Latin America, which inhibits model evaluation in these regions. Similarly, the lack of dry deposition comparison is due to the lack of measurements.

Y. Ge et al. (2021) also evaluate model outputs derived from two global emission inventories and find large localized discrepancies in NH₃, NO₂, and SO₂ emissions in certain world regions between the two inventories. The impact of these discrepancies on modelled concentrations varies with regard to primary or secondary species and differs regionally. Meanwhile, Y. Ge et al. (2021) indicate that the model shows generally better linear correlations with measurement networks in South East Asia ($\overline{R}$ = 0.73 over seven species), Europe ($\overline{R}$ = 0.67 over seven species), and North America ($\overline{R}$ = 0.63 over 7 species) than in China ($\overline{R}$ = 0.35 over 5 species), which implies potential discrepancies among different measurements or regional emissions rather than systematic issues with the model parameterization. The uncertainties in both model and measurement constrain the extent to which the agreement between model and measurement can be used to evaluate a model's performance. The values of statistical metrics for this EMEP–WRF system are at least as good as other global modelling studies (Bian et al., 2017; Hauglustaine et al., 2014). As already indicated, a detailed discussion of model and measurement uncertainties is presented in Y. Ge et al. (2021).

Additionally, the sensitivities of global and regional N_r, S_r, and PM_2.5 to various emissions reductions can only be investigated through modelling experiments, and since most model uncertainties will be similar across a set of simulations with the same model configuration, the modelled changes in concentrations between baseline and an emissions reduction scenario should be robust. Nevertheless, considering the fundamental uncertainties in emissions and model parameterizations, all numbers reported in this work should be considered as having underlying uncertainty, albeit that the latest available global emissions inventory and model version were used to minimize the impacts of these uncertainties.

4.2 Regional co-benefits and disbenefits of single-precursor emissions controls

The Results section shows that the reduction in emissions of one individual precursor has multiple co-benefits and sometimes small disbenefits on mitigating N_r, S_r, and PM_2.5 pollution, and these effects are geographically variable. In this work, our discussion focuses on East Asia, South Asia, Euro_Medi, and North America on account of both the thoroughly evaluated model results and the high population density plus high N_r and S_r pollution in these regions. The comparison of regional responses to emissions reductions reveals differences in regional oxidation regime, SIA chemistry, and deposition pattern, which are important processes to consider when designing emissions control policies, since transitory increases in PM_2.5 and some N_r and S_r pollutants could occur as emissions reduction measures are gradually applied.

Globally, reductions in NH₃ emissions are effective at reducing NH₃ concentrations and its wet and dry deposition but considerably less effective at reducing NH${}_{\mathrm{4}}^{+}$. This is because most world regions are in an ammonia-rich chemical domain, in which reducing NH₃ emissions only has limited effects on mitigating SIA formation (Ge et al., 2022). Other co-benefits of NH₃ emissions reductions include reductions in fine NO${}_{\mathrm{3}}^{-}$ surface concentrations and deposition in East Asia, South Asia, Euro_Medi, and North America because of reduced NH₄NO₃ formation. A notable disbenefit is the increased SO₂ surface concentration and human exposure in these regions, which is caused by reduced SO₂ dry deposition. The dry deposition velocity of SO₂ is negatively correlated with the molar acidity ratio a_SN, which is a model parameterization derived from long-term deposition measurements (Erisman et al., 2001; Simpson et al., 2012). Reduced NH₃ emissions therefore increase the acidity, leading to decreased SO₂ dry deposition.

Similarly, whilst reducing NO_x emissions is of course an effective way of decreasing global concentrations and deposition of OXN species, the degree to which different OXN species are decreased varies across regions. A 40 % reduction in NO_x emissions decreases NO_x and fine NO${}_{\mathrm{3}}^{-}$ surface concentrations in East Asia by 45 % and 33 %, respectively, whereas in South Asia this measure has a greater effect on fine NO${}_{\mathrm{3}}^{-}$ (45 % decrease) than on NO_x (22 % decrease). In Euro_Medi and North America, the 40 % NO_x emissions reductions produce similar decreases in regional average NO_x (36 %–38 %) and fine NO${}_{\mathrm{3}}^{-}$ (41 %–42 %) concentrations. These varying regional responses are consequent to different regional NO_x oxidation regime and SIA chemistry. The NO_x emissions reductions decrease NO_x surface concentrations, which increases O₃ concentrations in the high NO_x areas of eastern China and western and central Europe and therefore increases the atmospheric oxidizing capacity in these regions. As a result, more SO₂ is oxidized to SO${}_{\mathrm{4}}^{\mathrm{2}-}$, which leads to decreased SO₂ concentrations and deposition and, consequently, increased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations and deposition in these areas. The enhanced SO${}_{\mathrm{4}}^{\mathrm{2}-}$ production can partially (or even totally) offset the mass reduction in PM_2.5 caused by reduced NH₄NO₃ formation when reductions in NO_x emissions are not sufficiently high. The increased oxidant levels will also enhance HNO₃ and NO${}_{\mathrm{3}}^{-}$ production in these regions, but this effect does not compensate for the reduction in HNO₃ and NO${}_{\mathrm{3}}^{-}$ concentrations due to the reductions in their precursors (at least for 20 % and 40 % NO_x reductions), so the net effect is globally decreased HNO₃ and NO${}_{\mathrm{3}}^{-}$. Consequently, reduced HNO₃ and NO${}_{\mathrm{3}}^{-}$ levels caused by NO_x emissions reductions lead to less NH₄NO₃ formation, which then results in globally increased NH₃ concentration and deposition and decreased NH${}_{\mathrm{4}}^{+}$ concentration and deposition. In contrast, decreased O₃ concentrations in South Asia and North America in response to NO_x emissions reductions result in less chemical conversion of SO₂ to SO${}_{\mathrm{4}}^{\mathrm{2}-}$, which then causes increased SO₂ and decreased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations and deposition. Clappier et al. (2021) and Thunis et al. (2021) showed that the increased atmospheric oxidizing capacity induced by reductions in NO_x emissions is the reason for increased PM_2.5 levels in the Po basin (Italy), especially during wintertime, with increased nitrate, sulfate, and secondary organic aerosol (SOA) concentrations all being closely related to increased O₃ levels. Balamurugan et al. (2022) reported that reductions in SIA were much smaller than NO₂ emissions reductions during COVID lockdown in Germany, which is because the increased oxidant levels (OH, NO₃, and O₃) enhanced the formation of sulfate and nighttime nitrate, which then partially offset the lockdown-induced PM_2.5 decreases. Fu et al. (2020) noted that the increased oxidation of NO_x to HNO₃ due to increased O₃ levels makes winter haze NO${}_{\mathrm{3}}^{-}$ in the North China Plain (NCP) almost insensitive to 30 % reductions in emissions of NO_x, while Le et al. (2020) also revealed an unexpected PM exacerbation caused by unfavourable meteorological conditions and intensified SIA formation due to elevated O₃ levels induced by NO_x emissions reductions during COVID lockdown in China.

The greatest effect of SO_x emissions reductions is the direct decrease in global concentrations and deposition of SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}}$, which then induces changes in gas–aerosol partitioning of NH₃-NH${}_{\mathrm{4}}^{+}$ and HNO₃-NO${}_{\mathrm{3}}^{-}$. As discussed above, the reduction in (NH₄)₂SO₄ formation frees more gaseous NH₃ and promotes NH₄NO₃ formation, leading to increased concentrations and deposition of NH₃ and fine NO${}_{\mathrm{3}}^{-}$ in all world regions. Considering that one SO${}_{\mathrm{4}}^{\mathrm{2}-}$ takes up two NH₃ molecules under ammonia-rich conditions, but NO${}_{\mathrm{3}}^{-}$ only takes one, the net effect of SO_x emissions reductions still causes globally decreased NH${}_{\mathrm{4}}^{+}$ concentrations and deposition. Liu et al. (2018) noted a significant increase in annual NH₃ concentrations caused by rapid SO₂ emissions reductions in the NCP. In addition, the shifting of RDN from aerosol-phase NH${}_{\mathrm{4}}^{+}$ to gaseous NH₃ in response to reduced SO_x emissions also means that RDN pollution becomes more localized because NH₃ has a much shorter lifetime (1.6 d) than NH${}_{\mathrm{4}}^{+}$ (8.9 d), as it deposits more quickly to land rather than being transported to other regions (Ge et al., 2022). Utilizing combined measurements and modelling, Leung et al. (2020) found that the reduction in wintertime PM_2.5 in the NCP is buffered by enhanced NH₄NO₃ formation due to decreased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentration, liberating free NH₃, and increased oxidant levels, promoting HNO₃ production, despite SO₂ and NO_x emissions reductions in China. However, it is important to note that (NH₄)₂SO₄ has greater molecular mass than NH₄NO₃ and hence has larger leverage on PM_2.5 mass concentration, which ensures that the reductions in PM_2.5 mass concentrations derived from reduced (NH₄)₂SO₄ are less readily offset by increases in NH₄NO₃ concentrations. This also means that relative changes in the SO${}_{\mathrm{4}}^{\mathrm{2}-}$ component cause greater mass changes with respect to PM_2.5 air quality objectives (which must be expressed as mass concentration) than do the same relative changes in the NO${}_{\mathrm{3}}^{-}$ component. The relative changes in the SIA components of PM_2.5 expressed as molar concentrations would be different.

Additionally, it is noteworthy that changes in SO_x emissions have some subtle effects on NO_x concentrations as well. According to Fig. 4, small increases in NO_x concentrations are observed in southern China, western Europe, and the eastern US, while India and north-eastern China show decreased NO_x levels in response to SO_x emissions reductions. These impacts can be attributed to the role of SO${}_{\mathrm{4}}^{\mathrm{2}-}$ in the hydrolysis of N₂O₅ (which can undergo photolysis and produce NO₂). Many measurement studies, both laboratory and based on ambient samples, have revealed varying rates with different dependencies for this reaction, such as the relative humidity, available aerosol surface area, the size and composition of the aerosol particles, and the ratio of sulfate to organic matter (Bertram et al., 2009; Bertram and Thornton, 2009; Chang et al., 2011; Wagner et al., 2013; McDuffie et al., 2018). The parameterization used in the EMEP MSC-W model, based on the work of Riemer et al. (2003), incorporates SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (via the parameter of aerosol surface area and reaction probability) into the calculation of the hydrolysis rate of N₂O₅ (Simpson et al., 2012). Consequently, reductions in SO_x emissions affect SO${}_{\mathrm{4}}^{\mathrm{2}-}$ levels, which in turn affect N₂O₅ and NO_x levels. While the changes in NO_x concentrations are small when viewed at a regional average level (0 %–1 %, as depicted in Fig. 5), they may be significant in very local areas in the regions mentioned above. This part of the chemistry is one of the most uncertain aspects of the atmospheric science field. As a result, more laboratory and ambient measurement studies are needed to improve and test this type of model parametrization.

4.3 PM_2.5 sensitivities

This study highlights important messages for policy makers with respect to regional PM_2.5 mitigation. The most effective emissions control for decreasing regional average PM_2.5 concentrations, via an individual component, differs between world regions, as follows.

More emissions controls focusing on NH₃ and NO_x are necessary for regions with better air quality, such as Europe and North America, to decrease the annual PM_2.5 concentrations below the latest WHO guideline of 5 µg m⁻³ (WHO, 2021). More specifically, PM_2.5 sensitivities in Euro_Medi are complex and vary from the north to south. The UK and Scandinavia are more sensitive to NH₃ emissions reductions, central Europe is more sensitive to NO_x, while the Mediterranean region is more sensitive to SO_x, which is consistent with conclusions from other European studies (Megaritis et al., 2013; Vieno et al., 2016; Aksoyoglu et al., 2020; Jiang et al., 2020; Clappier et al., 2021). From a perspective of Europe-wide policy making, it is important to reduce NH₃ and NO_x emissions together and/or go for stronger reductions to minimize adverse effects caused by enhanced oxidation efficiency. In North America, the most effective measure to decrease annual mean PM_2.5 is reducing NO_x emissions. Similar conclusions are reported by other studies. Liao et al. (2008) reported that the reduction in NO_x emissions was most effective for decreasing 24 h mean PM_2.5 levels for five cities in the US, while Kelly et al. (2021) reported that reducing NO_x emissions was more effective for reducing PM_2.5 concentrations in the eastern US than SO₂, NH₃, and VOC emissions reductions in both January and July. Tsimpidi et al. (2007, 2008) showed that NO_x emissions reductions were the most effective measure for controlling PM_2.5 in the eastern US in summer due to the combined effects of lower atmospheric oxidant levels and smaller precursor emissions. Unfortunately, no conclusion on annual mean PM_2.5 sensitivities was drawn from these studies.

PM_2.5 in South Asia is most sensitive to SO_x emissions reductions and least sensitive to NH₃ emissions reductions, which is because South Asia is extremely ammonia rich (Ge et al., 2022), so reducing NH₃ has little impact on mitigating SIA. In this study, 40 % reductions in SO_x and NH₃ emissions, respectively, produce 10 % and 1 % decreases in the South Asia regional average PM_2.5. Pozzer et al. (2017) also showed that NH₃ emissions control has negligible impacts on PM_2.5 levels in South Asia, with a 50 % reduction in NH₃ emissions only reducing annual mean PM_2.5 by 2 %. Given the continuing strong growth of SO₂ emissions in India now and in future projections (Sadavarte and Venkataraman, 2014; Hoesly et al., 2018; Aas et al., 2019; Szopa et al., 2021), it is anticipated that PM_2.5 levels in this region will continue to rise.

For East Asia, although considerable emissions reductions in SO₂ have been taking place in China, and to a lesser extent reductions in NO_x (Liu et al., 2016; Hoesly et al., 2018; Zheng et al., 2018), continuous efforts need to be put into reducing not only SO₂ and NO_x but also NH₃, as reductions in the three precursors are equally effective for mitigating PM_2.5. Studies focusing on East Asia also highlight the importance of NH₃ emissions control for reducing annual PM_2.5 pollution, especially in central and eastern China (Wang et al., 2011; Pozzer et al., 2017; Cheng et al., 2021). In addition, due to the high proportion of SIA in PM_2.5 in East Asia, reducing all three precursors together also produces the greatest reduction in regional average PM_2.5 compared to South Asia, Europe, and North America. Another benefit of collective emissions reductions is that it minimizes the adverse side effects from individual reductions, since these disbenefits can cause transitory increases in PM_2.5 and deposition of certain N_r and S_r species when emission reduction measures are gradually applied.

In short, considering the evolution of the chemical state of the atmosphere in Europe and North America since the industrial revolution, the measures to combat PM_2.5 pollution in the two regions may be a good reference for countries in East and South Asia.

4.4 Non-one-to-one linearity

Analyses of both 20 % and 40 % emissions reductions reveal different kinds of linearity and non-linearity. There is considerable non-linearity in the sense of a lack of one-to-one proportionality between an emission reduction and a species concentration or deposition change. Precursor emissions reductions sometimes even increases other pollutant concentrations. This is consequent on interactions in SIA formation, atmospheric oxidizing capacity, and N and S deposition, as discussed earlier, and is highly geographically variable. Such non-one-to-one proportionality may be more significant in a certain area during a certain season due to the inter-annual variability in local meteorology and emissions profiles. However, whilst the sensitivity of an annual quantity of a pollutant to emissions reductions is subject to seasonality, we are confident that the broad conclusions of this study will hold, and from a global and regional policymaking perspective, it is more practical to develop policies of reductions in emissions from different sectors and countries on an annual level. Studies focusing on non-proportional responses of European PM_2.5 to emissions reductions showed that significant seasonality only occurs in a few specific areas (Thunis et al., 2015, 2021; Clappier et al., 2021).

On the other hand, a linearity in response to emissions reductions is apparent via the observation that the responses of PM_2.5, N_r, and S_r annual concentrations and deposition components remain essentially proportional to the precursor emissions reductions (20 % and 40 %) for a given precursor in a given region, albeit that the magnitude of the slope varies substantially with different precursors and regions. Even if the net concentration changes of one species (e.g. PM_2.5) induced by reductions in emissions of all three precursors are smaller (or greater) than the sum of changes from reductions in individual precursors due to different chemical interactions as discussed earlier, these net changes still follow a very similar gradient when emissions reductions in all three precursors change from 20 % to 40 %. However, where the gradient in the response is not one-to-one, the linearity in the response that is observed up to the 40 % emissions reductions simulated here clearly cannot continue to extrapolate linearly all the way to 100 % emissions reductions. The gradient of the response must be flatter or steeper (depending on atmospheric component) at the beginning or end of the span from 0 to 100 % emissions reduction. For instance, 20 % and 40 % reductions in NH₃ emissions give 24 % and 48 % decreases in global annual mean NH₃ concentrations, respectively (i.e. gradients exceeding one to one), whereas these emissions reductions only produce 6 % and 14 % decreases in global NH${}_{\mathrm{4}}^{+}$ concentrations, respectively (i.e. gradients less than one to one). When NH₃ emissions are completely switched off, both NH₃ and NH${}_{\mathrm{4}}^{+}$ concentrations will be zero, so the NH₃ concentration sensitivity must become smaller, and the NH${}_{\mathrm{4}}^{+}$ concentration sensitivity must become larger as NH₃ emissions reductions approach 100 %. Additional serial sensitivity experiments are required to acquire the full spectrum of NH₃ and NH${}_{\mathrm{4}}^{+}$ (and all other species) sensitivities. It is therefore important for policy makers in different regions to know the emissions reductions required to obtain the mitigation responses needed for specific air quality targets.

4.5 Implications and limitations

The parameterization of atmospheric processes in ACTMs is in continuous interplay with the experimental evidence. The responses of N_r and S_r species to emissions reductions in NH₃, NO_x, and SO_x as discussed in previous sections reveal several complex processes incorporated into the EMEP MSC-W model. The extent to which these parameterized processes accurately represent the reality of the atmosphere calls for more experimental evidence (field and/or laboratory measurements).

For instance, the synergistic interaction between NH₃ and SO₂ dry deposition to leaf surfaces is well known and has been termed “co-deposition”. The idea is that the existence of both NH₃ and SO₂ acts to reduce the canopy resistances and increase the dry deposition rate for both gases (Sutton et al., 1994; Fowler et al., 2001; Nemitz et al., 2004; Fowler et al., 2009; Massad et al., 2010). This co-deposition effect is incorporated into the EMEP MSC-W model, making use of empirical parameters derived from European field measurements (Erisman et al., 2001) which may not be completely applicable to other world regions (e.g. South America). In addition, for regional variations in atmospheric oxidizing capacity caused by NO_x emissions reductions, there are several studies reporting measurement evidence in East Asia, Europe, and North America, while no such measurements are found in South Asia. Long-term standardized measurements for both surface concentrations and wet and dry deposition with a sufficient amount of sampling sites in each geographical region are required for corroborating globally modelled results. The quantitative results reported in this study can serve as a reference for future modelling and measurement studies, albeit firmer field evidence worldwide is required to develop a reliable and robust chemistry and deposition scheme for global models.

Finally, it is also important to remember that reductions in anthropogenic emissions of SIA precursors will have many co-benefits on forest health, ecosystem biodiversity, and climate, not just in populated areas but elsewhere. For instance, NH₃ has become a major air pollution driver of lichen distributions in many European forests in recent years. In Scotland, Sutton et al. (2009) showed how lichens were gradually eradicated in areas near a poultry farm which is a large emitter of NH₃. Similarly, van Herk (2001) reported that increased ambient NH₃ concentrations in the Netherlands appear to be the primary cause of the disappearance of acidophytic lichen species (i.e. species that prefer naturally acidic bark) over the last decade. Moreover, although the effects on the availability of nutrients in terrestrial and aquatic ecosystems are broadly assumed to be decided by total N inputs, Sutton et al. (2020) placed a stronger emphasis on the form in which the N was deposited. In their experiments, dry deposition of NH₃ showed a larger toxicity than wet deposition of NH${}_{\mathrm{4}}^{+}$ and NO${}_{\mathrm{3}}^{-}$. In this case, policy makers should be more cautious about emissions controls with side effects of increased NH₃ dry deposition. Meanwhile, since most NH₃ is emitted via volatilization, a warmer atmosphere will promote its global emissions (Johnson et al., 2008; Sutton et al., 2013; Riddick et al., 2018). NH₃ emission controls thus need to include both direct reductions and indirect measures to abate climate warming as well.

5 Conclusions

The sensitivities of global and regional annual mean surface concentrations and deposition of gaseous and particle N_r and S_r to 20 % and 40 % reductions in anthropogenic emissions of NH₃, NO_x, and SO_x, both individually and collectively, has been investigated using the EMEP MSC-W model coupled with WRF meteorology for 2015. East Asia, South Asia, Euro_Medi, and North America are selected for regional discussions because of their high population densities and N_r and S_r pollution, and because the model outputs are evaluated and agree reasonably well with measurements in most of these areas. The comparison in regional responses reveals that the emissions reduction in one precursor has multiple co-benefits and sometimes small disbenefits on mitigating N_r, S_r, and PM_2.5 pollution, and these effects are highly geographically variable.

Whilst reductions in NH₃ emissions are effective for decreasing annual NH₃ concentrations and deposition, they are considerably less effective at decreasing NH${}_{\mathrm{4}}^{+}$. This is because all densely populated continents are ammonia rich, so reducing NH₃ emissions only has limited effects on mitigating SIA formation. A 40 % reduction in NH₃ emissions decreases regional average NH₃ concentrations in the four regions by 47 %–49 %, while NH${}_{\mathrm{4}}^{+}$ concentrations decrease in the order of Euro_Medi (18 %), East Asia (15 %), North America (12 %), and South Asia (4 %), the order of increasing regional ammonia richness. A notable disbenefit is increased SO₂ concentrations because reduced NH₃ levels affect the pH-dependent SO₂ dry deposition. A 40 % reduction in NH₃ emissions increases SO₂ concentrations in East Asia by 16 %, in South Asia and North America by 14 %, and in Euro_Medi by 10 %.

Large regional differences are observed in NO_x emissions reduction scenarios. In East Asia, NO_x concentrations are very effectively decreased (by 45 %) with 40 % NO_x emissions reductions, but they are less effectively decreased in Euro_Medi (38 %) and North America (36 %), and to the least extent in South Asia (22 %). By contrast, the regional sensitivities of fine NO${}_{\mathrm{3}}^{-}$ are reversed: South Asia shows the largest decrease (45 %), whilst East Asia shows the smallest decrease (33 %). This phenomenon is related to different regional oxidation regimes and SIA chemistry. NO_x emissions reductions increase O₃ levels in East Asia (and also, but by less, in Euro_Medi) but decrease O₃ levels in South Asia (and also, but by less, in North America), which causes an enhanced NO_x and SO_x oxidation in former regions but a decreased one in latter regions. Consequently, increased SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and SO₂ concentrations appear in East Asia and South Asia, respectively.

Reductions in SO_x emissions have globally consistent impacts on SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations. A 40 % reduction in SO_x emissions decreases SO₂ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in the four regions by 42 %–45 % and 34 %–38 %, respectively, while the disbenefit is that decreased (NH₄)₂SO₄ formation yields ∼ 12 % growth in NH₃ total deposition.

This work also highlights important messages for policy makers concerning the mitigation of PM_2.5. More emissions controls focusing on NH₃ and NO_x are necessary for regions with better air quality, such as Europe and North America. In Euro_Medi, PM_2.5 sensitivities vary from the north to south, with NH₃ reductions being more effective for the UK and Scandinavia, NO_x for central Europe, and SO_x for the Mediterranean. In South Asia, PM_2.5 is most sensitive to SO_x and least sensitive to NH₃, which is consistent with the fact that South Asia is so ammonia rich that reducing NH₃ hardly has any impacts. Given the continuing strong growth of SO₂ emissions in India now and in future projections, it is anticipated that PM_2.5 levels in this region will continue to rise. In East Asia, although considerable emissions reductions in SO₂ have been taking place in China, and to a lesser extent reductions in NO_x also, continuing efforts need to be put into reducing not only SO₂ and NO_x but also NH₃, as reductions in the three precursors are equally effective for mitigating PM_2.5.

This work reveals some geographically varying and non-one-to-one proportionality of chemical responses of N_r, S_r, and PM_2.5 to emissions reductions. It is thus important not only to prioritize different emission controls in different regions but also to reduce several emissions together in order to minimize the potential disbenefits.