The sensitivity test to adjust the CASTNET-measured concentrations based on the reported average differences between weekly-duration and 24 h duration samples shows little difference between the unadjusted and adjusted pH values in the USA (2.69 ± 0.85 and 2.74 ± 0.83 on average for the unadjusted and adjusted pH, respectively), suggesting that the weekly duration of the CASTNET sampling has little impact on the calculated aerosol pH. Therefore, we proceed with our subsequent analyses using the unadjusted pH. The aerosol pH values calculated based on observational data show a significant difference between China (most observation sites are in the NCP) and the USA. In China (mainly the NCP), the 2017 annual average pH at monitoring sites is 4.3, ranging from 3.3 to 5.4 with an interquartile range of 3.9–4.6. In the contiguous USA, the 2011 annual average pH is 2.6, ranging from 1.9 to 3.9 with an interquartile range of 2.2–3.0 (Fig. 1). The t test shows a statistically significant difference between the two groups (p< 0.0001), suggesting that the aerosols are on average more acidic at the monitoring sites in the contiguous USA than in China (NCP).

The pH difference is also illustrated by the cumulative distribution function (CDF) curves (Fig. 2, solid lines). The shapes of the CDF curves are similar in these two countries with a slightly steeper slope in the contiguous USA (Fig. 2a). The pH values, however, are 1–2 units higher in China (NCP) than in the contiguous USA across varying levels of cumulative frequencies in the CDF curves. In some cases, aerosols could be completely neutral in China (NCP) (the frequency is 2 % for pH ≥ 7), while in the contiguous USA, the pH values in all cases were below 6.

Spatially, 14 out of the 16 sampling sites in China are in the NCP (Fig. S2c) which is one of the most populous and polluted regions in China (Hu et al., 2014; Cui et al., 2020). Our pH results in this region are consistent with those of other studies (ranging from 3.5 to 4.6) (Liu et al., 2017; Ding et al., 2019; Ge et al., 2019). The distribution of sampling sites in the USA, on the other hand, is more evenly distributed spatially. The pH values in the midwest and California are higher than in other regions like the southeast, in line with previous studies (Lawal et al., 2018; Chen et al., 2019). Overall, the pH level in the USA is 1.7 units lower than over the NCP of China.

To address the issue of insufficient spatial coverage of the observational data in China, we conduct simulations using CMAQ, in conjunction with the observational data, to further study the pH difference on a nationwide scale. We evaluate the model performance by comparing the modeled and observed aerosol pH values (Fig. 3); major particle and gaseous species including SO42-, NO3-, NH4+ and HNO3, and NH3; and the partitioning ratios including ε(NH4+) and ε(NO3-), at monitoring sites (Figs. S4–S6).

Spatially, the model simulations generally capture the observed variations in pH, species concentrations, and partitioning ratios, although there are some notable biases (Figs. S4 and S5). In both China (NCP) and the contiguous USA, the modeled NH4+, NO3-, and NH3 are biased low while modeled HNO3 is biased high, resulting in low biases in the predicted ε(NO3-) and ε(NH4+). The modeled SO42- in both countries is biased low. Such low biases have been seen in previous studies (Fountoukis et al., 2013; Theobald et al., 2016) and have been attributed to the spatial mismatch between the observations and simulations due to the coarse resolutions of model grid cells (usually 20–50 km resolution) (Shen et al., 2014; R. Wang et al., 2014). Smaller NMBs in the USA indicate a better performance compared to China (NCP). Larger differences between observations and simulations in China (NCP) could also be caused by larger measurement uncertainties as the data in China are collected from different monitoring stations operated by individual research institutions (Wang et al., 2019) and thus lack unified quality control compared with data in the USA, which come from national monitoring networks (United States Environmental Protection Agency, National Atmospheric Deposition Program). The co-occurrence of low biases in ε(NO3-), which causes lower bias in aerosol pH, and low biases in ε(NH4+) and SO42-, which cause higher bias in aerosol pH, likely offset each other, resulting in small biases in aerosol pH. Indeed, the simulated average pH values at observation sites (3.8 ± 0.2 in the NCP, China, and 1.8 ± 0.5 in the contiguous USA) are generally in line with the observed averages (4.3 ± 0.5 in the NCP, China, and 2.6 ± 0.5 in the contiguous USA) (Fig. 3), although the model shows a moderate low bias in both countries. The larger pH difference in the USA than in China is likely due to the low bias in TNH3 to which the sensitivity of pH is found to be more pronounced in the USA than in China (discussed in detail in Sect. S1).

With respect to the temporal variation, the model captures the seasonal trends of pH, ε(NH4+), and ε(NO3-) in both countries, with lower values in summer and higher values in winter (Fig. 4). The lower temperature in winter favors the partitioning toward the particle phase for semi-volatile species. Comparison of the seasonal trends of the individual aerosol components shows a better agreement in the USA than in China. For example, the simulation in the USA captures the trends of almost all components, though it is biased low for SO42- and NH4+ in summer (Fig. S6b and h); the simulation in China misses the peaks of SO42- in winter and NH3 in summer and has high biases for HNO3 in summer (Fig. S6a, i, and e). Measurement-related biases may contribute to the disparity in the temporal trends between observed and modeled concentrations. The uncertainty in monthly profiles of emission estimates may also play an important role. For example, CASTNET's long sampling period could lead to a larger measurement bias in summer than in winter (Sickles and Shadwick, 2008); the large uncertainty in the current estimates of NH3 emissions in China, especially the reported underestimation of summertime emissions as indicated by an inversion analysis (Kong et al., 2019), may cause the absence of the summertime NH3 peak in the simulated trend (Fig. S6i). Further investigation is needed to better understand the factors underpinning the disparity between observations and model simulations. In spite of the various potential uncertainties, overall, the spatial and temporal evaluation suggests generally good agreement between the model simulations and observations in both countries.

In line with observations (Sect. 3.1.1), the nationwide model simulations show significant differences in aerosol acidity between the two countries. Almost all the areas in the USA have aerosol pH values lower than 3 according to the CDF (Fig. 2b). Higher pH values are found in the middle and eastern USA, while in the western USA except California, the pH values are lower (Fig. 3). In China, a large portion of areas (87 %) have aerosol pH values above 3 according to the CDF. This is especially true in eastern China which has the largest population (Fig. 3). Aerosol pH values in western and southeastern China are generally lower than in the east. It should be noted that due to the scarcity of observational data, the pH estimates in southern and western China are not evaluated. The nationwide annual average pH values in China and the USA are 2.7 ± 0.6 and 0.8 ± 0.8 units, respectively, lower than the observation-based values due partly to the model bias but also because most of the monitoring sites are in areas with high pH (Fig. 3).

Given the adverse health impacts of ambient aerosols (Burnett et al., 2014; Freedman et al., 2019) and the potential linkage of aerosol acidity with aerosol toxicity through the solubility of redox-active metals (Oakes et al., 2012; Fang et al., 2015; Ye et al., 2018), we further calculate and compare the population-weighted averages of aerosol pH in the two countries to highlight the pH levels in densely populated areas. The population-weighted pH values are 3.3 ± 0.4 and 2.2 ± 0.5 in China and the USA, respectively, both of which are higher than non-weighted averages, indicating that aerosols in more populous areas tend to be less acidic (Fig. 2b). This finding is further confirmed by the statistically significant positive correlation (α=0.01), within each country, between the aerosol pH and population density (for China, r=0.42 and p< 0.0001; for the USA, r=0.28 and p< 0.0001). Consistent with the observation-based results, the t test for the model simulations shows a significant difference in both the population-weighted and non-weighted aerosol pH values between the two countries (p< 0.001).

We further investigate the factors leading to the pH difference. Although both observations and simulations are subject to uncertainty, we expect observations to provide more direct and reliable evidence for this investigation. It should be noted that the monitoring sites in China were clustered in the NCP and, thus, may not be representative of the whole of China. Table 1 summarizes the annual average concentrations of gaseous and particle species measured in China (NCP) and the contiguous USA during the study period (China, 2017; USA, 2011). For all the gaseous and ionic species (except HNO3), the average concentrations in China (NCP) are statistically significantly higher than those in the contiguous USA. The total concentrations of WSI species in China (NCP) (34.4 µg m-3) are on average 6 times the concentrations in the contiguous USA (5.7 µg m-3) and have greater variation, ranging from 0.2–240 µg m-3, compared to a range of 0.1–31 µg m-3 in the contiguous USA. Similarly to in other studies in China (Yao et al., 2002; Pathak et al., 2009; Zhang et al., 2013; Liu et al., 2016) and the USA (Guo et al., 2015; Feng et al., 2020), NH4+, NO3-, and SO42-, contribute more than 80 % of the total WSI concentrations in both countries. The mass fractions of individual WSIs, however, differ between the two countries (Fig. 5). In China (NCP), the dominant WSI was NO3- (34.6 %), followed by SO42- (26.3 %) and NH4+ (25.5 %). In the contiguous USA in 2011, SO42- contributed nearly half of the total WSI concentration (49.4 %), and the contributions of NO3- and NH4+ are comparable (NO3- 17.6 %, NH4+ 18.8 %). Note that SO42- and NO3- levels have been decreasing dramatically over the years, leading to decreases in NH4+ since there is less substrate to interact with NH3 and form particulate ammonium species (Butler et al., 2016).

Two of the most predominant anions in aerosols, SO42- and NO3-, at the monitoring sites in China (NCP) are present at 4 and 15 times the concentrations, respectively, of those observed in the contiguous USA. The relative difference in NO3- between the two countries is the most significant, compared with the differences in other WSI components. Hence, the difference in the nitrate-to-sulfate molar ratio (NO3- / SO42-) is also significant between the two countries. The observational data show that the ratios at most monitoring sites in China (NCP) are larger than 1 and that only two sites have ratios lower than but close to 1 (0.81, 0.94). On the other hand, 27 out of the 34 sites in the contiguous USA show a ratio lower than 1, ranging from 0.25–0.99. High NO3- / SO42- in China (NCP) could be caused by more efficient oxidation of NOx than SO2 in China leading to greater nitrate formation as well as higher aerosol pH and availability of NH3 which favor the formation of particle nitrate (Guo et al., 2018b; Vasilakos et al., 2018). The varying ratios of NO3- / SO42- in aerosols could further affect aerosol liquid water uptake, which is discussed in the Supplement (Sect. S2).

The most abundant cation in aerosols is NH4+, and the concentration difference in NH4+ between the two countries (by a factor of 11) is more significant than the difference in other cations (by factors of 2–7). In addition, ε(NH4+) in China (NCP) (0.13–0.48) is approximately 50 % lower than in the contiguous USA (0.22–0.85), meaning that compared to the USA, TNH3 in China tends to be present more in the gas phase. Higher NH4+ and lower ε(NH4+) levels in China amount to a higher level of TNH3 which has an important influence on aerosol pH, partitioning of TNO3, and even particulate mass (see Supplement for more discussion).

NVCs such as Na+, Ca2+, Mg2+, and K+ are often minor components of particles but are important because of their ability to neutralize acidic species in the atmosphere, such as sulfuric and nitric acids (Zhang et al., 2007). Neglecting NVCs would cause low biases in pH, driving the NH3–NH4+ equilibrium to shift toward the particle phase because more ammonium would be used to neutralize the aerosols that would otherwise be neutralized by NVCs (Guo et al., 2018a). Therefore, NVCs are included in calculating aerosol pH in this study. High NVC concentrations usually occur at the sites near emission sources. For example, high concentrations of Na+, mainly from sea salt (Zhang et al., 2011), occur at sites 13, 27, and 30 in the USA, which are all coastal sites. High concentrations of Ca2+, mainly from mineral dust, are found at sites 6, 11, and 23 in the contiguous USA and at Site 5 in China (NCP), which are in prairies impacted by sand and dust. Average NVC concentrations in China (NCP) are up to an order of magnitude higher than in the contiguous USA, although in both countries, most of the NVCs concentrations are small compared to those of SO42-, NO3-, and NH4+. The predominant NVCs in China (NCP) are Ca2+ (2.8 %), K+ (2.1 %), and Na+ (2.0 %), and in the contiguous USA, they are Ca2+ (5.9 %) and Na+ (3.7 %).

We use the MTSM as described in Sect. 2.4 to characterize the contribution of each component to the pH difference between the USA and China. Three groups (i.e., observation, simulation non-weighted, simulation population-weighted) of the annual average concentrations in the USA and China listed in Table S4 are chosen as the starting (USA) and ending (China) points to perform the analysis. The results are shown in Fig. 6.

The average concentrations based on the observational and simulated data are not completely consistent due to the representativeness of the monitoring sites and the discrepancy between the simulations and observations. The MTSM analyses based on the three groups, however, show similar results. For example, all three groups suggest the high TNH3 in China as an important factor leading to the difference in aerosol pH between the two countries (Fig. 6). The contribution of TNH3 is the highest in the observation group due to the large difference in TNH3 concentration. Other cations, mainly NVCs, have a relatively small effect (0.2, 0.2, and 0.3 in groups “observation”, “simulation”, and “simulation-weighted”, respectively), which is consistent with a previous study (Zheng et al., 2020). Unlike TNH3 and NVCs which lead to higher pH values in China than in the USA, SO42- contributes in the opposite direction to the pH difference. High SO42- concentrations decrease aerosol pH in China by 0.6–1.3 units, compared to the USA, although this effect is fully offset by TNH3.

Compared to other species, the concentrations of TNO3 are the most different between the two countries (by a factor of 15), but the MTSM shows that the contribution of TNO3 to the pH difference is relatively small (0.1, 0.1, and 0.2 in the observation, non-weighted, and population-weighted groups). This result is further confirmed by a sensitivity test of TNO3 (Fig. S10) which shows that the change in pH from changing only TNO3 is small in both countries. More detailed analyses and discussions on the effects of TNH3, TNO3, and SO4 on aerosol pH can be found in the Supplement.

Studies have identified an important role of temperature in driving aerosol pH (Battaglia et al., 2017; Tao and Murphy, 2019; Jia et al., 2020). Our MTSM analysis shows that temperature accounts for 0.07–0.39 units of pH difference between China and the USA, which varies by group (Fig. 6). Such relatively small contributions of temperature, compared to those of TNH3 and SO4, are mainly because of the small difference in temperature between these two countries which are at similar latitudes. The difference in the annual average temperature between China and the USA is 1.4, -5.0, and 2.6 K in the observation, non-weighted, and population-weighted groups, respectively (Table S4).

As aerosol pH is calculated as [log10(LWC) - log10(Hair+) - 3], all mechanisms affecting aerosol pH must be through the modification of LWC, Hair+, or both (LWC and Hair+ are expressed as mass per unit volume of air, µg m-3). We quantitatively separate the contributions of individual factors to the China–US pH difference into the LWC-modifying pathway and the Hair+-modifying pathway (Fig. 6). To achieve this, we use the MTSM to quantify the contributions of individual factors to the differences in log10(LWC) and [-log10(Hair+)-3] between the two countries, with the same approach as we did for pH (LWC and Hair+ are two output variables directly predicted by ISORROPIA). The results show that the changes in both LWC and Hair+ lead to increases in aerosol pH when conditions change from those in the USA to those in China.

Given that LWC increases with aerosol mass concentration (Song et al., 2019), higher component concentrations in China than in the USA increase LWC and, thus, increase aerosol pH (Fig. 6b). Through the LWC-modifying pathway, changes in SO4, TNH3, and TNO3 lead to increases in pH (0.15–0.3) (Fig. 6b), which are consistent in all three groups. Compared to other groups, the observation group represents a higher pH increase due to Cl and a higher pH decrease due to RH (Fig. 6b), mainly because of the larger differences in Cl concentrations and RH for this group than for other groups (Table S4).

Through the Hair+-modifying pathway, the effects of individual factors on pH changes vary (Fig. 6c). Increases in acidic components (SO4 and TNO3) increase Hair+ and thus decrease aerosol pH (Fig. 6c). Increases in TNH3, TCl, and NVCs, on the other hand, decrease Hair+ and increase aerosol pH (Fig. 6c). By increasing Hair+, increased SO4 decreases pH by 0.7–1.2 units, showing a much stronger acidic capacity than another acidic component, TNO3, which only decreases pH by 0.17–0.27 units (Fig. 6c). Compared to the USA, China is in a TNH3-rich condition. The molar ratios of [TNH3]/(2×[SO4]+[TNO3]+[TCl]) in China vs. in the USA are 3 vs. 1.4, 2.0 vs. 1.0, and 2.4 vs. 1.5 in the observation, non-weighted, and population-weighted groups, respectively. Changing the conditions from the USA to China, TNH3 plays the most important role in neutralizing the acidic components and driving the pH increase in the Hair+-modifying pathway (Fig. 6c).

For individual factors, the net changes in pH are a result of the combination of the two pathways. For example, increased SO4 increases LWC and Hair+ simultaneously. The increase in LWC increases aerosol pH, while the increase in Hair+ decreases aerosol pH. All three groups suggest that the effect of Hair+ on pH overwhelms that of LWC on pH, leading to a net decrease in pH from an SO4 increase (Fig. 6). Increased TNH3 increases pH in both pathways, adding up to a larger increase in pH (Fig. 6). Increased TNO3 through these two pathways, however, is associated with opposite effects on pH which are comparable in magnitude and thus tend to offset each other (especially in the observation group) (Fig. 6). This explains the aforementioned small change in pH from the TNO3 increases. Combining all the factors, both pathways increase aerosol pH (Fig. 6b and c), resulting in the large difference in aerosol acidity between these two countries (Fig. 6a).

To facilitate a follow-up sensitivity test to link the two pathways with mass concentration and chemical composition, we define the total mass concentration as the sum of the mass concentrations of all the eight input components (i.e., Na+, SO4, TNH3, TNO3, TCl, Ca2+, K+, and Mg2+), including both gas and particle phases, and the chemical composition as the composition of the eight components in the aerosol (gas + particle) system. The observation group shows that the total mass concentration in China is 8.4 times that in the USA, and the chemical composition in China is richer in TNH3 than that in the USA (as illustrated by the ratios of [TNH3]/(2×[SO4]+[TNO3]+[TCl]) mentioned above). It has been found that both LWC and Hair+ are affected by mass concentration and aerosol composition (Guo et al., 2015; Zheng et al., 2020; Xie et al., 2020). To investigate how the differences in mass concentration and composition between China and the USA are associated with the LWC- and Hair+-modifying pathways and consequently the pH difference, we first increase the mass concentrations of individual input components in the US case by a constant factor of 8.4, whereby we obtain an intervening case representing the overall pollution level as in China but with the chemical composition feature as in the USA (Table S4, sensitivity test). From the intervening case, we then shift the composition of the US case to that of China (Table S4, sensitivity test). Note that throughout this sensitivity test, meteorological conditions are held constant. The first step, by increasing the mass concentration, yields an increase of 1.02 units in the aerosol pH, which is mainly achieved through the LWC-modifying pathway (1.06 units) instead of through the Hair+-modifying pathway (-0.04 units) (Fig. S7a–c). The second step that changes the chemical composition shows a further increase of 0.76 units in the aerosol pH, which is mainly achieved through the Hair+-modifying pathway (0.88 units). The LWC-modifying pathway plays a minor role (-0.11 units) in this step (Fig. S7d–f). This sensitivity test reveals that the LWC-modifying pathway is mainly associated with the change in mass concentration and the Hair+-modifying pathway is mainly associated with the change in chemical composition.

It is surprising that in the first step, pH changed when the concentrations of all chemical components were scaled by a common factor. This means that pH changes with mass concentration of the aerosols (gas + particle) even though all chemical component mole fractions hold. Further investigation shows that increasing the aerosol concentration drives TNO3 and TNH3 partitioning toward particle phases – ε(NH4+) and ε(NO3-) increase from 0.4 and 0.6 to 0.6 and 0.98, respectively. Given the weak acidity of NO3-, the particle is ultimately neutralized by the increased NH4+. The repartitioning in response to the increase in mass concentration is thus key to the pH shift and can be explained by Henry's law; i.e., [Aaq]=HA⋅pA, where [Aaq] is the aqueous-phase concentration of component A in units of moles per liter of water, pA is the partial pressure of A in the gas phase, and HA is Henry's law coefficient (Seinfeld and Pandis, 2006). [Aaq] is proportional to cA/LWC (cA denotes the particle-phase concentration of A; note that LWC and cA are expressed in mass per unit volume of air and [Aaq] is expressed in moles per unit volume of water). Increasing the concentrations of all chemical components by a common factor increases pA (due to the increase in the gas-phase concentration of A) but does not change [Aaq] (because both cA and LWC increases in the same direction by the same magnitude). According to Henry's law, more gas-phase A will thus shift toward the particle phase to achieve a thermodynamic equilibrium of the new system.

We find that by increasing the concentration of every component by a constant factor, the magnitude and direction of the resulting change in pH are sensitive to the fraction of TNH3 in the aerosol system while insensitive to the ratio of SO4 to TNO3. Based on an NH4+–SO42-–NO3-–H2O system, we conduct a series of sensitivity tests to investigate the change in aerosol pH in response to the multiplication of a constant factor of 8.4 (Fig. S8). The change in pH reduces gradually from 1.2 units to 0.8 units when the TNH3 mass fraction of the system decreases from 67 % to 27 % (Fig. S8). With further decreases in the TNH3 fraction, the increase in pH diminishes rapidly, becomes negative when the TNH3 mass fraction is lower than 25 %, and is -0.6 when the TNH3 mass fraction is 17 % (Fig. S8). Under a constant TNH3 mass fraction, the change in pH remains generally constant across a wide range of the mass ratios of SO4 to TNO3 (from 5 : 1 to 1 : 5) (Fig. S8). In populated continental regions, mass fractions of TNH3 are often high (Bencs et al., 2008; Behera and Sharma, 2010; Zheng et al., 2015; Cheng et al., 2016; Guo et al., 2017b), and an increase in mass concentration thus typically increases the aerosol pH.

Such an assessment by tracking pathway- and step-specific contributions provides a better understanding of the pH difference between China and the USA. We show that through the LWC-modifying pathway, the increases in aerosol components consistently lead to increases in pH and that through the Hair+-modifying pathway, the effects of different components on pH vary in direction. If the LWC-modifying pathway dominated the pH changes over the Hair+-modifying pathway, aerosol mass concentrations would be the main factor driving the aerosol acidity difference between China and the USA and one could simply attribute the difference in aerosol acidity to the fact that China is more polluted than the USA. In contrast, if the Hair+-modifying pathway dominated, chemical composition would be the dominant factor and the compound profiles of precursors emissions, which affect the fractions of the corresponding aerosol components in the air, would play an important role. While there has been debate about whether mass concentration or chemical composition plays a more important role in determining aerosol pH (Cheng et al., 2016; Guo et al., 2017a; Pye et al., 2020; Zheng et al., 2020), our results suggest that both are important in explaining the China–US pH difference (Fig. 6b and c). The three groups are not consistent with each other in which pathway contributes more than the other to the pH difference, but they all suggest that the two pathways are comparable in terms of their effects on aerosol pH (Fig. 6b and c).

Our results, showing the importance of both mass concentrations associated with LWC and chemical composition associated with Hair+ and a minor role of temperature, seem in some aspects to contradict a previous study (Zheng et al., 2020) which highlighted LWC and temperature instead of chemical composition as the most important factors explaining the pH difference between China (NCP) and the USA. We note that the difference in the conclusions is reasonable when considering the differences in the specific cases examined in these two studies. The previous study compared the conditions in NCP in winter with those in the southeastern USA in summer (SE USA). Because of the differences in latitude (north for China vs. south for the USA) and season (winter for China vs. summer for the USA), the difference in temperature between their scenarios (29 K) was an order of magnitude greater than those in our study, which has greater spatial and temporal coverage (2.6 K in the observation group, 5 K in the non-weighted group, and -1.4 K in the population-weighted group). Using the MTSM, we evaluate the pH difference between the NCP and SE-USA scenarios considered in the previous study. The results show that temperature accounts for 1.3 units of difference in aerosol pH between their two scenarios (Fig. S9), in line with what was previously reported (1.6 units).

In addition, ISORROPIA simulations show a LWC difference of 8.2 µg m-3 between China (NCP) and the contiguous USA in the observation group in our study and 340 µg m-3 between the scenarios considered in the previous study. The much larger LWC difference in the previous study compared to ours is mainly driven by the differences in pollutant concentrations. For example, the SO4 concentration is as high as 156 µg m-3 in the NCP scenario in the previous study but only 9.2 µg m-3 in our study. Such differences in concentrations are reasonable, given that the previous study selected a severe haze event occurring in Beijing in winter 2013 as the scenario for China (NCP), while we use annual average levels over the NCP in 2017 as our case for China (NCP). Note that winter 2013 was a period when air pollution reportedly reached record high levels across northern China (Y. Wang et al., 2014; Li et al., 2016). Since 2013, China has launched strict controls on air pollutant emissions, and PM2.5 levels decreased significantly between 2013 and 2017 (Zhang et al., 2019). Therefore, the NCP scenario in the previous study should be more representative of short-term haze events in the pre-2013 period, while our China (NCP) case should be more representative of annual average levels in recent years.