The impact of European legislative and technology measures to reduce air pollutants on air quality, human health and climate

We used the composition-climate model HadGEM3-UKCA, as described in Turnock et al (2015), to simulate surface aerosol mass concentrations as well as the impact on aerosol REs. HadGEM3-UKCA is similar in complexity to other models that have participated in the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP, (Lamarque et al 2013)) and the AeroCom initiative (Mann et al 2014). Simulations had a horizontal resolution of 1.875° by 1.275° (∼140 km at mid latitudes). Our previous work has demonstrated that HadGEM3-UKCA is able to capture the regional changes in annual mean surface aerosol mass concentrations and surface solar radiation observed over Europe in response to changes in anthropogenic emissions during the period 1980–2010 (Turnock et al 2015). Our simulations therefore capture regional changes in PM_2.5 (at the scale of ∼100 km) and are suitable for isolating the effects of policy driven changes in regional aerosol precursor emissions. Our model underestimates PM_2.5 concentrations in urban areas; we estimate the impact of this in section 2.3. Further details on the model description and experimental set-up are in supplementary methods section 1.

Monthly mean anthropogenic emissions of CO, SO₂, NO_X, NMVOC, OC and BC are based on version v4.3 of the Emission Database for Global Atmospheric Research (EDGAR) (EC-JRC/PBL 2015). The emissions used in this study are described in Crippa et al (2015) and are summarised below. Emission fields are provided at 0.5° by 0.5° resolution for the energy, industry, transport, agriculture, residential, waste and other sources sectors (http://edgar.jrc.ec.europa.eu/pegasos).

We compare two simulations with different anthropogenic emissions. All other factors in the model are the same. The reference simulation (REF) uses actual 2010 emissions. The NO-AQ simulation uses anthropogenic emissions that would have occurred in 2010 if measures to reduce air pollutant emissions between 1970 and 2010 did not occur. The NO-AQ scenario is based on 2010 economic activity data but assumes 1970 emission factors to simulate the stagnation of technological improvements in the energy, industry and road transport sectors. It also assumes that no further end-of-pipe abatement measures (e.g., particle filter traps, fuel desulphurisation) occurred in the energy and road transportation sectors after 1970 (Crippa et al 2015). The NO-AQ scenario does however account for changes in the makeup of the national energy supply (e.g. coal to gas) that occurred without specific legislation. These assumptions alter emissions globally, since stagnation of technology improvements are applied globally and some European emission standards are used outside Europe (Crippa et al 2016). However, we focus on the impact of these measures over Europe (defined here as the domain 10°W–36°E and 36°N–61°N).

Figures 1(a)–(c) shows the spatial distribution of the difference in emissions between the two scenarios (REF minus NO-AQ). This difference represents the emissions that were averted in 2010 due to the implementation of technology improvements and end-of-pipe abatement measures. The largest difference in emissions occurs over central and eastern Europe and the UK. Total European SO₂, BC and OC emissions over all sectors are reduced, compared to what would have occurred, by 9.4 Tg SO₂ yr⁻¹ (53%), 0.4 Tg BC yr⁻¹ (59%) and 0.3 Tg OC yr⁻¹ (32%) (figure 1(d)). Reductions in SO₂ emissions are dominated by changes in the energy sector, whereas reductions in BC and OC are dominated by changes in the transport sector (table S1).

Zoom In Zoom Out Reset image size

Figure 1(d) compares total European emissions in the NO-AQ scenario against the actual 2010 (REF) and 1970 emissions from the EDGAR and MACCity inventories. The emissions of SO₂ and OC in both the 2010 REF and NO-AQ scenarios are lower than emissions in 1970. In contrast, BC emissions from the 2010 NO-AQ scenario are larger than in 1970 but are lower in the 2010 REF scenario. This suggests that in the absence of technological improvements and end-of-pipe abatement measures, emissions of SO₂ and OC would still have declined substantially due to changes in the makeup of the energy supply and consumption rates whereas emissions of BC would have been larger than in 1970. This highlights the large impact of legislation and technological improvements on reducing BC emissions compared to SO₂ and OC and that historical changes to anthropogenic emissions are determined by the combined effect of legislation, technology and economic factors.

Concentration response functions (CRFs) for long-term exposure to PM_2.5 have been used here to quantify the adverse health effects in terms of excess premature mortality from ischemic heart disease, stroke, chronic obstructive pulmonary disease, lung cancer and acute lower respiratory infection in infants. Here, the link between PM_2.5 concentrations and relative risk of the above diseases is based on the integrated exposure-response functions developed by Burnett et al (2014), which also uses the baseline rate of disease and spatial distribution of the population.

The resolution of global aerosol models means that they typically underestimate urban PM_2.5 concentrations. To estimate the impact of this underestimation on simulated human health impacts, we complete another calculation where we increase the PM_2.5 concentrations in European urban areas in both the REF and NO-AQ simulations. We then apply the CRFs to these modified concentrations. We estimate that the underprediction of PM_2.5 in urban areas in our global model (Turnock et al 2015) causes an bias in our mortality estimates of 13%–16%, substantially less than the uncertainty in the CRF (see supplementary methods). Our method is outlined fully in supplementary methods section 2.

The impact on human health from poor air quality can also have a perceived economic effect on society. The methods and data within WHO Regional Office for Europe (2015) have been used to calculate the perceived economic benefit to society that result from the health benefits associated with the reduction in particulate matter concentrations due to the implementation of air pollution mitigation measures (see supplementary methods section 3).

Based on the climate response to regional radiative forcing (Shindell and Faluvegi 2009), and the atmospheric energy budget (Andrews et al 2010, Richardson et al 2015), we estimate the temperature and precipitation response due to changes in the radiative balance from the implementation of air pollutant (particulate matter) mitigation measures (see supplementary methods section 4).

Figure 2 shows the simulated change in surface annual mean PM_2.5 concentrations due to the implementation of European air pollutant emission reduction measures (REF versus NO-AQ simulations). European average concentrations are calculated as the 5 year annual mean (±standard deviation) over the continental land mass within the domain 10°W–36°E and 36°N–61°N. Across Europe, annual mean PM_2.5 concentrations reduce by 1.75 ± 0.04 μg m⁻³ (35%), sulphate by 1.1 ± 0.03 μg m⁻³ (44%), BC by 0.27 ± 0.01 μg m⁻³ (56%) and particulate organic matter (POM) by 0.36 ± 0.01 μg m⁻³ (23%).

Zoom In Zoom Out Reset image size

Reductions in PM_2.5 mass concentration of 40%–50% occur across much of central, eastern and parts of southern Europe (figure 2(a)). Sulphate mass concentrations decrease by more than 50% across most of central Europe and south eastern Europe (figure 2(b)), mainly due to the substantial decrease in SO₂ emissions in these regions (figure 1(a)) that mostly occurs in the energy sector (table S1). The largest absolute reduction in sulphate mass concentrations (of >5 μg m⁻³) occurs in south-eastern Europe (figure S1). This is attributed to large reductions in SO₂ emissions from the energy sector combined with atmospheric conditions that enhance the efficiency of sulphate formation (higher oxidant concentrations and temperatures than northern Europe). Reductions in BC mass concentrations of more than 70% occur across all of western Europe and Scandinavia (figure 2(c)), mainly due emissions reductions over Europe from the road transport sector (table S1). Fractional reductions in POM mass concentrations are typically less than 40% (figure 2(d)), despite a similar absolute reduction compared to BC (figure S1). The smaller fractional change in POM concentrations are due to the larger background concentrations of POM from natural emissions, which are the same in both simulations.

Legislative and technology measures have therefore led to reduced aerosol mass concentrations across Europe, with potential impacts for both regional climate and human health. We compared the change in aerosol concentration calculated due to these air quality mitigation measures against the change in aerosol concentrations that actually occurred over the period 1970–2009 simulated by Turnock et al (2015) using the same model. Using this method we estimate that air pollutant mitigation measures are responsible for 29% of the actual reductions in PM_2.5 that occurred over the same period. Air pollutant mitigation contributes the most to changes in BC concentrations (73%) compared to sulphate (33%) or POM (15%). This difference reflects the different causes behind the changes in anthropogenic emissions and highlights the important effect of enacted European legislation on air quality.

Ammonium nitrate aerosol (absent from our simulations) has increased by 15%–30% over Europe since the 1980s due to the reduction in SO₂ and sulphate aerosols (Fagerli and Aas 2008). The increase in ammonium nitrate aerosol concentrations would partially offset some of our simulated improvements to air quality and health but would not alter the overall net decrease in European mean particulate matter concentrations (1.75 μg m⁻³) due to the larger declines in sulphate (44%), BC (56%) and POM (23%) simulated here.

Figure 3 shows the impacts of improvements in regional air quality from air pollutant mitigation measures on public health. The simulated reduction in PM_2.5 concentrations due to air pollutant mitigation measures reduces the number of premature deaths across Europe, with the largest numbers prevented across central Europe (Germany and The Netherlands) (figure 3(a)). Air pollutant mitigation measures prevent 3 to 4 premature deaths annually per 10 000 people in central and eastern Europe (figure 3(b)) and 5 to 6 premature deaths annually per 10 000 people in south eastern Europe (Romania and Bulgaria), where the largest reductions in PM_2.5 occurred (figures 2(a) and S1(a)).

Zoom In Zoom Out Reset image size

Across all the European Union (EU), reductions in PM_2.5 due to air pollutant mitigation measures prevented 80 250 (37 380–115 840, at 95% confidence intervals of the CRFs) premature deaths annually (individual country totals on figure 3(c) and table S2), which is 19% of the estimated premature deaths due to present-day (2011) PM_2.5 concentrations across Europe.

We compared the improved public health due to these air quality mitigation measures against an estimate of the improved public health due to the actual change in PM_2.5 concentrations that occurred over the same period. Using the change in simulated PM_2.5 concentrations between 1970–74 and 2005–2009 from Turnock et al (2015), we estimate that 210 000 annual premature deaths were prevented over this period. Therefore European policies to reduce air pollutants and technological improvements can be considered responsible for ∼38% of the total number of premature deaths prevented due to the improved air quality across the EU from 1970 to 2009 (figure 3(c)).

We estimate that the improvement in human health from implementing air pollutant mitigation measures results in a perceived economic benefit to society of US$232 billion across the EU (individual country totals on figure 3(d) and table S3), 1.4% of the EU's GDP in 2010. In particular, the perceived economic benefits to society for some countries in eastern Europe, such as Hungary and Bulgaria, are relatively more important and represent a greater percentage of their national GDP (>5%). The WHO calculated that in 2010, premature mortality due to ambient concentrations of particulate matter was costing the EU US$700 billion, or 4% of the EU's GDP (WHO Regional Office for Europe 2015). We therefore suggest that the perceived economic costs of poor air quality would be 33% greater than they are currently if European legalisation and technological improvements had not been implemented.

Table 1 shows the difference in European annual mean aerosol RE at the TOA and at the surface due to the implementation of the air pollutant emission mitigation measures. The reduction of aerosol concentrations in the atmosphere over Europe (figure 2) causes a positive change in the aerosol RE at the TOA, reduces the atmospheric absorption of radiation (mainly due to reductions in BC) and also increases the shortwave radiation incident at the surface. The annual European mean all-sky aerosol RE at the TOA increases by 2.3 ± 0.06 W m⁻² with clear-sky TOA radiation increasing by 1.7 ± 0.05 W m⁻². The shortwave radiation incident at the surface increases under both all-sky (3.3 ± 0.07 W m⁻²) and clear-sky conditions (2.7 ± 0.05 W m⁻²). Air pollutant mitigation measures have reduced atmospheric absorption of radiation over Europe by ∼1 W m⁻² in both all-sky and clear-sky conditions (table 1 and figure 4) due to the large reductions in BC aerosols. Our model shows that air pollution mitigation has resulted in a warming effect on climate compared to what would have occurred, dominated by changes in the direct aerosol RE (clear-sky). This occurs from a combination of warming due to reductions in scattering aerosols (sulphate and POM) and the cooling response from the reduction in absorbing BC aerosols.

Zoom In Zoom Out Reset image size

We can compare the change in aerosol RE calculated here against an estimate of the RE that actually occurred over the period 1970–2009 (table 1). In transient simulations with historical emissions, Turnock et al (2015) estimated that the annual European mean all-sky aerosol RE increased by 3.1 W m⁻² between 1970 and 2009. The larger change in RE over the period 1970–2009 reflects the similar changes in BC but larger reductions in sulphate and POM that occurred in reality. Therefore our model simulations show that European legislation and technological improvements to reduce air pollutants are responsible for approximately 74% of the change in all-sky TOA aerosol RE from the 1970s to present day.

Over the 20th Century nitrate aerosols (not present within our model) were shown by Bellouin et al (2011) to have had a relatively small impact on the negative global aerosol forcing, enhancing it by less than 20%. The calculated changes in aerosol REs in this study will have been overestimated due to the absence of the negative aerosol REs from ammonium nitrate.

Figure 4 shows the spatial pattern of changes to the aerosol RE at the TOA and at the surface. The largest increase in TOA clear-sky aerosol RE occurs over south-east and central Europe (figure 4(b)), corresponding with the largest reduction in sulphate mass (figures 2(b) and S1). The largest reductions in atmospheric absorption (TOA minus surface) occur over The Netherlands and Germany where the largest reductions in BC aerosols occurs (figure 2(c)). The TOA warming response over these areas would have been even larger without the reductions of the absorbing BC aerosols (cooling response).

The all-sky changes are also caused by changes to cloud properties (aci). Air pollution mitigation reduces concentrations of cloud condensation nuclei (CCN) (figure S3(a)) leading to larger, less numerous cloud droplets (figure S3(b)), increasing shortwave all-sky radiation at the TOA and surface (figures 4(a) and (c)). Air pollutant mitigation causes the largest increase in all-sky radiation (both TOA and surface) over coastal areas of western Europe (figures 4(a) and (c)), corresponding with the largest reduction in CCN, cloud droplet number concentrations and larger cloud fractions (figures S3 and S4). Over central Europe, CCN and cloud drop concentrations are larger (shown by larger total number concentrations in figure S2), meaning air pollutant mitigation causes smaller fractional changes and a smaller RE here. The difference between all-sky and clear-sky radiation changes (figure S3(d)) gives an indication of the cloudy sky RE, which is greatest over north west Europe.

Overall reductions in aerosols from the implementation of air pollutant mitigation measures resulted in a positive change in the aerosol RE at the TOA and also increased the amount of solar radiation incident at the surface under both all-sky and clear-sky conditions. The simulated change in European all-sky TOA aerosol RE due to air pollutant mitigation is 2.5 times the change in global mean CO₂ radiative forcing over the period 1970–2009 (Myhre et al 2013), indicating the large regional impact of air pollutant mitigation measures. Such changes to the radiative balance over Europe could have non-negligible effects on the climate, impacting both surface temperature and precipitation patterns.

Using the change in TOA aerosol RE between the two simulations, we estimate that the reduction in particulate matter from the implementation of air pollutant mitigation measures have increased European mean surface temperatures by 0.45 ± 0.11 °C compared to temperatures that would have occurred without mitigation measures (for methods see section 4 of supplementary methods). This change in European surface temperature is anticipated to be slightly smaller than that presented here due to the absence of ammonium nitrate from the simulations. In addition, the reduced atmospheric absorption of radiation caused by mitigation measures will induce a rapid adjustment to increase European precipitation by 13 ± 0.8 mm yr⁻¹. Future work is needed to contrast the human health and economic costs associated with these changes in climate against those quantified here due to air quality improvements.