The impact of climate mitigation measures on near term climate forcers

Here we quantify the regional co-benefits to future air quality on annual to daily mean timescales from implementing mitigation measures to stabilise future climate. Two consistent future emissions pathways are used within the composition-climate model HadGEM3-UKCA: one is a reference pathway of future economic growth and development (REF), whilst the Representative Concentration Pathway 4.5 (RCP4.5) assumes the same development pathway but stabilises anthropogenic radiative forcing at 4.5 W m−2 in 2100. Implementing greenhouse gas (GHG) mitigation measures in RCP4.5 reduces global mean air pollutant emissions by up to 30% in the 2050s, in addition to mitigating climate. Annual mean surface concentrations of ozone and PM2.5 decrease by 10%–20% from the combined reductions in emissions and climate change. The number of days exceeding the World Health Organization’s (WHO) daily mean air quality standards are reduced by up 47 days for ozone and 15 days for PM2.5 over different world regions. The air quality co-benefits from mitigation measures are mainly achieved from reductions in anthropogenic emissions, although benefits can be offset due to changes in climate. In terms of anthropogenic climate forcing, while the reduction in global mean effective radiative forcing (ERF) in 2050, relative to the 2000s, due to enacting carbon dioxide mitigation measures (−0.43 W m−2) is enhanced by decreases in tropospheric ozone (−0.26 W m−2) and methane (−0.2 W m−2), it is partially offset by a positive aerosol ERF from reductions in aerosols (+0.35 W m−2). This study demonstrates that policies to mitigate climate change have added co-benefits for global and regional air quality on annual to daily timescales. Furthermore, the effectiveness of the GHG policies in reducing anthropogenic climate forcing is enhanced in the near-term by reductions in ozone and methane despite the increased forcing due to reductions in aerosols.


Introduction
The air pollutants ozone (O 3 ) and particulate matter (PM) can have a detrimental impact on human health (Lelieveld et al 2015) and variable impacts on ecosystems (Fowler et al 2009). Additionally, both pollutants are 'Near Term Climate Forcers' (NTCFs) because they influence climate in the short-term (due to their short atmospheric lifetime) by perturbing the Earth's radiative balance (Myhre et al 2013). O 3 acts as a greenhouse gas (GHG) whereas PM both scatters and absorbs radiation (aerosol radiation interactionsari), in addition to altering the microphysical properties of clouds (aerosol cloud interactions-aci). Changes to climate (via meteorological parameters) can also affect the spatial distribution and concentrations of air pollutants (von Schneidemesser et al 2015, Doherty et al 2017, Silva et al 2017. Future mitigation measures targeted at reducing air pollutant emissions are generally enacted to improve local air quality and benefit human health (US EPA 2011, Turnock et al 2016. However, policies focussed on the mitigation of climate, through the reduction of GHGs, can also inadvertently impact the concentration and spatial distribution of air pollutants in the atmosphere (von Schneidemesser et al 2015). This can occur through changes in the rate of air pollutants co-emitted from carbon sources and by changes in the physical climate. The effect of future changes in climate on PM is uncertain (Jacob andWinner 2009, Allen et al 2016) but climate change is generally considered to decrease background O 3 concentrations (Isaksen et al 2009, Fiore et al 2012 and worsen surface O 3 in polluted regions-the so-called 'climate penalty' (Rasmussen et al 2013, Colette et al 2015. The number of premature mortalities associated with exposure to PM 2.5 (particles with a diameter less than 2.5 micrometres) and O 3 is likely to increase under future climate change (Doherty et al 2017, Silva et al 2017. Less stringent climate mitigation measures, leading to a larger temperature response, could also eliminate any future benefits to surface O 3 from reductions in precursor emissions (Fortems-Cheiney et al 2017). It is therefore important to consider the impact from air quality and climate mitigation measures together as future air quality will be determined by the combined effect of both.
Model studies using the Representative Concentration Pathways (RCPs) in the 5th Coupled Model Intercomparison Project (CMIP5) simulated both positive and negative changes to surface O 3 and PM concentrations in the 2050s, relative to the 2000s, from the combined effects of changes in emissions and climate (Kirtman et al 2013, Young et al 2013, Kim et al 2015. For O 3 , reductions in global mean surface concentrations are simulated in the 2050s for all RCPs apart from RCP8.5 whereas, global mean PM 2.5 concentrations are predicted to decrease in all scenarios. Reductions in surface O 3 and PM 2.5 concentrations of up to 20% in 2050 were simulated over Europe, Asia and North America using future climate mitigation measures targeting mainly methane (CH 4 ) and black carbon (BC) sources (Stohl et al 2015). However, there is a large range of regional responses (positive and negative) in O 3 and PM, both within models and to the different future pathways, highlighting the large uncertainty in future estimates (Fiore et al 2012). Changes to future PM 2.5 concentrations tend to be smaller than for O 3 due to the larger contributions from natural sources (e.g. dust, sea-salt emissions), that are inherently more variable due to the assumed future climate state.
Solely implementing climate mitigation policies has been shown to improve future air quality and human health at a carbon price that is less than air pollution abatement costs (Shindell et al 2012, West et al 2013, Vandyck et al 2018. Results from multiple Integrated Assessment Models (IAMs) on the co-benefits of climate policies for air quality found a benefit to crop yields and a reduction in global exposure to PM 2.5 above World Health Organization (WHO) values in 2050, but with large variations over India and Africa (Rao et al 2016, Vandyck et al 2018. Combining climate and air quality mitigation measures results in 39% of the global population in 2050 being exposed to PM 2.5 concentrations less than the WHO annual guideline value of 10 μg m −3 (Rao et al 2016). Applying carbon dioxide (CO 2 ) mitigation measures over the USA in the 2050s reduces surface concentrations of O 3 and PM 2.5 and causes a negative O 3 and positive aerosol radiative forcing (Lee et al 2016, Zhang et al 2016. Over Europe, implementing climate policies provides cost savings from reduced health impacts in 2050 due to reductions in both PM 2.5 (mainly emission driven changes) and O 3 , although the sign of the O 3 response depends on the magnitude of future climate change (Schucht et al 2015). Over China, simulating an illustrative climate policy for peaking CO 2 emissions in 2030 reduces annual mean PM 2.5 concentrations by up to 12%, with 94 000 associated avoided premature mortalities and associated co-benefits that are larger than mitigation costs (Li et al 2018). Air quality co-benefits for health and crop yields could be larger if the ambitious long-term goals of the Paris Agreement on climate change are met (Reis et al 2018, Vandyck et al 2018. These studies highlight the potential co-benefits to air quality and human health that can be achieved from solely implementing climate mitigation measures.
It is important to further understand the impact from climate mitigation measures on air pollutants, as uncertainties still exist in the future impact on air quality and climate, especially the extent to which they could alter the efficacy of air quality mitigation measures. Previous studies have tended to only focus on the air quality and health co-benefits, whereas here we systematically assess the impact of climate mitigation measures on regional air pollutants (O 3 and PM 2.5 ) across different timescales (annual, seasonal, daily) and on near-term climate forcing. In this study we use a set of self-consistent future climate scenarios (RCPs used in CMIP5), in a similar way to West et al (2013), but within a fully coupled global composition-climate model (HadGEM3-UKCA). Here we perform a systematic assessment of the impact on air pollutants (O 3 and PM 2.5 ) from climate mitigation measures across specific world regions on daily, seasonal and annual timescales, to assess the exposure of the world's population to concentrations above WHO guideline values. The influence from the reduction in co-emitted sources and climate change are quantified separately by using additional sensitivity simulations. By analysing 30 years of global climate simulations, that include interactive chemistry and aerosols coupled to climate, we are able to more fully consider the effect of climate variability on the future projection of air pollutants (Garcia-Menendez et al 2017, Shen et al 2017). This is particularly important for Earth system feedbacks, which can be strongly affected by different future climate states (e.g. changes to natural emission sources of aerosols, and have not been previously considered in this context). We also quantify for the first time the additional climate co-benefit from reducing carbon emissions on the change in effective radiative forcing (ERF) from the near-term climate forcers relevant to air quality; O 3 , CH 4 , BC and total PM. The air pollution emission controls assumed in REF are the same as in RCP4.5, allowing any differences in emissions to be attributable to the climate mitigation policies within RCP4.5. The magnitude of change in carbon emissions in 2050 between RCP4.5 and REF is a mid-range climate policy scenario, intermediate between the Nationally Determined Contributions and reductions required to meet the 2°C goal set out in the Paris Agreement. The REF to RCP4.5 change in SO 2 and NO X , key drivers of air pollution, is also of intermediate magnitude when compared to changes amongst the shared socio-economic pathways (SSPs) used in CMIP6 between the reference and 4.5 W m −2 scenarios (Rao et al 2017). Whereas, changes in other air pollutants (CO, NMVOCs, BC and OC) between RCP4.5 to REF are larger than in the SSPs (figure S1). Figure 1 and table S1 show that implementing climate mitigation measures, the difference between RCP4.5 and REF (referred to as RCP4.5 2050 andREF 2050), has a relatively large impact on regional air pollutant emissions in 2050. A global reduction in air pollutant emissions of up to 30% in the 2050s results from solely implementing the climate mitigation measures within RCP4.5, with larger regional reductions. Carbon monoxide (CO) and BC emissions are reduced by more than 50% over North America and Russia, attributed to a reduction in biomass burning of the boreal forest. For some species (e.g. NOx, SO 2 ), the reduction in air pollutant emissions over Europe and North America from solely implementing climate mitigation measures in the 2050s, tends to be smaller than that from combined air pollution controls and climate policies over the period 2000to 2050(RCP4.5 2050-BASE 2000. However, over the rapidly developing regions of Africa and South Asia, implementing climate mitigation measures could offset some of the anticipated increase in air pollutant emissions e.g. SO 2 .

Model set up
In this study, an atmosphere only configuration of the fully coupled HadGEM3-UKCA composition climate model (see section S2 of the supplementary for details) was used at a horizontal resolution of 1.875°by 1.275°( ∼140 km at mid latitudes) and 85 vertical levels (up to 85 km). Timeslice simulations were conducted where the climate in each scenario is represented by using decadal mean prescribed sea ice (SI) distributions, sea surface temperatures ( Tables 1 and  S2 summarise all simulations performed in this study, which have been conducted for a 30 year averaging period to more fully account for the influence of climate variability.
The difference between the RCP4.5 2050 and REF 2050 simulations will show the impact on future air quality from undertaking mitigation measures to stabilise climate. The magnitude of the reduction in carbon emissions and changes in air pollutants considered here in 2050 are from a mid-range climate policy scenario (figure S1), meaning that the overall air quality co-benefits will be intermediate. A student t-test has been performed to assess whether the difference is significant at the 95% confidence interval. To isolate the influence of changes in climate versus that from co-emission sources, a simulation (REF4. 5 2050 A summary of the surface O 3 and PM 2.5 model evaluation, in terms of the spread in normalised mean bias factors at observation sites, is shown in figure 2 for different timescales (annual, seasonal and daily) and across different regions. The model slightly overpredicts annual mean surface O 3 concentrations in the northern hemisphere (by a factor of 1-1.5) and underpredicts southern hemisphere concentrations (by a factor of ∼1.5), in a similar way to other global models (Young et al 2018). Seasonally, the model underestimates wintertime and overestimates summertime surface O 3 measurements. Figure 2 shows that Had-GEM3-UKCA generally underpredicts annual mean surface PM 2.5 concentrations, with a better representation of observed summertime values and a consistent low model bias in wintertime. The model observational biases for surface PM 2.5 are similar to those identified before (Turnock et al 2015) and in other global and regional models (Glotfelty et al 2017,  .5 , the model under-represents the elevated daily surface concentrations during pollution episodes, which is expected when using a global model driven by decadal mean monthly emissions at a horizontal resolution of >100 km. The magnitude of seasonal concentrations and similar daily events in the future will also be underestimated by the model, although the simulated change between future scenarios will be consistent, but conservative in magnitude.

Results and discussion
3.1. Regional impact on air quality  (table S4), which is strongly driven by the reduction of co-emitted tropospheric O 3 precursors (−2.6 ppbv) but partially offset by the effect of climate change (temperature reduction) acting to increase background (i.e. non-episodic) surface O 3 concentrations (+0.4 ppbv) due a reduction water vapour and O 3 loss (table S5). Largest benefits to surface O 3 occur in spring and summer over most northern hemisphere regions. Maximum reductions of up to 3 ppbv occur in JJA over Europe, East Asia and North America, mostly due to the reduction of coemitted tropospheric O 3 precursors (<−3 ppbv). Changes to climate from mitigation measures tend to increase surface O 3 concentrations over most regions by less than 1 ppbv. Surface O 3 increases by 0.3 ppbv over the Pacific, Australia and New Zealand region in DJF due to changes in climate (+0.4 ppbv) with a slight offset from emission changes (−0.1 ppbv).
Seasonal mean changes in population weighted surface PM 2.5 concentrations due to implementing climate mitigation measures are smaller and more variable (figure 4). Global mean population weighted PM 2.5 concentrations decrease by 0.6 -1.2 μg m −3 in most seasons, mainly due to the reduction of sulphate, BC and organic matter from emission changes (figure S3). However there is an increase in global mean population weighted PM 2.5 concentrations in March, April, and May (MAM: +0.2 μg m −3 ), which is dominated by changes in dust and organic aerosol sources (figure S3). Climate change mitigation measures act to increase PM 2.5 (+1.0 μg m −3 , figure S5) in MAM, which outweigh benefits from emission reductions (−0.8 μg m −3 , figure S4). A large reduction in population weighted surface PM 2.5 concentrations of >2 μg m −3 occurs over South Asia in DJF from emission reduction measures, primarily from SO 2 decreases (figure 1). Certain regions (Central Asia, Middle East and North Africa) are influenced by natural PM (dust) sources and exhibit a large variability in their PM 2.5 response. However, the large overlapping error bars on figure 4 and absence of stippling on figures S3-S5 show that these changes in PM 2.5 near the dust source regions are not significant. Future projections of surface PM 2.5 concentrations are more variable and harder to attribute to a particular influence, as indicated by the large overlapping error bars from emission and climate drivers.
Overall, implementing GHG mitigation measures reduces global population weighted annual mean surface concentrations of O 3 and PM 2.5 in 2050 by 1.6 ppbv and 0.5 μg m −3 , respectively. Benefits and penalties to regional air pollutants from measures to stabilise climate are shown to vary both regionally and seasonally. The global and regional co-benefits to air quality are smaller in this study for O 3 and PM 2.5 and have a more variable response for PM 2.5 than in West et al (2013) and Zhang et al (2017). The difference between studies can be attributed to the use of a different model and the larger simulated influence of climate change on ozone (globally 15%) and PM 2.5 (globally 30%). The increase in surface O 3 solely due to the effects of climate change is spatially consistent with that in West et al (2013) but slightly larger due to the higher climate sensitivity of the model (HadGEM2; Collins et al 2011) used to provide SST and sea ice November. Stippling shows differences that are statistically significant at the 95% confidence interval using a student t-test.
fields (Andrews et al 2012). The influence of climate variability was shown to be particularly important for interactive natural sources of aerosols (dust and sea salt) (figures S4 and S5). It is therefore important to account for climate variability in model simulations where Earth System feedbacks could limit improvements from future anthropogenic emission controls. Nevertheless, the benefits from co-emission reductions outweigh any penalties from mitigating climate and result in a net benefit of climate policies to surface O 3 and PM 2.5 concentrations, in accordance with West et al (2013) and Zhang et al (2017).  figure 5 and table S6). Largest benefits occur across Asia, Europe and North America where the regional mean exceedance reduces by more than 25 days (table S6). The reductions in the 2050s over these regions are even larger when compared to the number of exceedances in the 2000s. However, for regions like Africa, Middle East and South Asia climate mitigation measures act to limit any future increase in the number of exceedances compared to the 2000s.

Impact on WHO air quality guideline values
Whilst the absolute magnitude of simulated daily mean PM 2.5 concentrations is underestimated by the model (figure 2), there is a change in the number of daily exceedances due to climate mitigation measures, which is smaller than that for O 3 due to the influence of climate variability on emissions from natural sources (section 3.1). Changes in PM 2.5 from dust source regions contributes to the small increase in the number of days that daily mean PM 2.5 exceeds the WHO AQGV over the Central America, South America and Pacific, Australia, New Zealand regions. Across the anthropogenic source regions of East Asia and South Asia, climate mitigation measures reduced the regional mean exceedances of the WHO AQGV by more than 10 days (table S6). Compared to the large number of exceedances in the 2000s, climate measures provide an additional reduction of exceedances for East Asia and limits future increases over South Asia.
In summary, climate mitigation measures provide an additional reduction in the daily exposure of the population to elevated surface concentrations of both O 3 and PM 2.5 , particularly over regions (e.g. South and shown in solid coloured bars). The regional difference due to solely emission related changes (RCP4.5 2050-REF4.5 2050) is shown by the X symbol, whereas that due to changes solely in climate (REF4.5 2050-REF 2050 is shown by+symbol. The error bars on the X and+symbols represent the standard deviation in the mean regional response over the 30 year simulation period. East Asia) where there is currently a high population exposure.

Changes to ERF
As both O 3 and PM are NTCFs, any change in their concentration could have an impact on climate forcing in addition to air quality. Implementing the climate mitigation measures in RCP4.5 not only reduces CO 2 and CH 4 concentrations and their radiative forcing compared to REF but will also inadvertently change O 3 and aerosol (PM) radiative forcing.

Conclusions
Mitigation measures to stabilise future climate are targeted at reducing emissions of CO 2 , its radiative forcing, and future climate change. However, implementing these measures has the potential to have an inadvertent impact on concentrations of air pollutants from changes in future climate, as well as to co-emitted precursors. Here we use a coupled compositionclimate model to assess the impact of future climate mitigation measures on air pollutants by using simulations with the same air pollutant controls but different climate policies. Globally, climate policies reduce the co-emission of air pollutants in 2050 by 10%-30%, in addition to stabilising climate at a lower value of global mean surface temperature. From our simulations we estimate that the implementation of climate mitigation measures reduces surface concentrations of O 3 and PM 2.5 , with annual mean co-benefits of up to 10% regionally. Larger benefits of up to 15% are simulated in summertime for surface O 3 and in wintertime for PM 2.5 over northern hemisphere anthropogenic source regions. Scenarios used in this study were those used in CMIP5 as newer scenarios, providing a greater number of future trajectories with differing levels or air pollutant and climate mitigation, were not available (Rao et al 2017, Gidden et al 2019. The extent of climate mitigation considered here in 2050 represents a mid-range climate policy scenario, with air quality co-benefits that are of intermediate magnitude ( figure S1). The scenarios used in CMIP6 provide a larger range of future climate and air pollutant emission trajectories, with the potential for larger or smaller air quality co-benefits.
The co-benefit to surface air quality is mainly achieved through the reduction of co-emitted air pollutants and their precursors, in accordance with West et al (2013). However, over certain regions, there are small increases in air pollutants in response to the changes in climate induced from climate mitigation measures. Simulated co-benefits of climate mitigation measures to air quality are smaller here than in West et al (2013) due to the use of a different model and the larger simulated influence of climate change. This highlights the need for further investigation (through a multi-model assessment) on the importance of Earth System feedbacks in limiting any benefits from future emission controls.
Implementing climate mitigation measures reduces the exposure of the population to daily concentrations of PM 2.5 and O 3 above the WHO air quality guideline values. For regions currently experiencing high levels of air pollution, such as East and South Asia, there are notable reductions in the number of days that concentrations of PM 2.5 and O 3 exceed the WHO Air Quality Guideline Values in the 2050s. Climate mitigation measures are therefore able to provide an additional reduction, on top of direct air pollutant controls, in the daily exposure of the population to high levels of air pollutants, with associated co-benefits for human health (e.g. West et al 2013).
O 3 and PM, are radiatively active and can influence the Earth's radiative balance as well as impact on surface air quality. Changes to their concentrations from implementing climate mitigation measures alters their radiative effect on climate. Reductions in aerosols (PM) increases their radiative forcing in 2050 (relative to 2000) whilst reductions in O 3 reduce its radiative forcing. Reductions in radiative forcing from CO 2 , CH 4 and O 3 due to climate mitigation measures benefits anthropogenic climate forcing, although this is offset by the increased aerosol forcing.
Future implementation of climate mitigation measures results in a co-benefit to both air quality and climate. Surface concentrations of air pollutants are reduced, mainly from decreases in co-emitted precursors, which improves air quality and reduces the daily exposure of the population to high concentrations. Anthropogenic climate forcing is reduced due to decreased CO 2 , CH 4 and tropospheric O 3 precursors. However, this benefit is partially offset by the reduction in PM causing an unintended positive (warming) forcing to climate. It is therefore important to consider both the air quality and climate impact from nearterm climate forcers in any future climate and air quality policies.