A new set of historical and future anthropogenic air pollutant emissions have been developed and used as part of CMIP6. The historical anthropogenic emissions are from the Community Emissions Data System (CEDS), and a new dataset was developed for biomass-burning emissions, both of which provide information on emissions from 1750 to 2014 (van Marle et al., 2017; Hoesly et al., 2018). The SSPs used in future CMIP6 experiments represent an update from the RCPs used in CMIP5 as they combine pathways of socio-economic development with targets to achieve a certain level of climate mitigation (O'Neill et al., 2014; van Vuuren et al., 2014; Riahi et al., 2017). The SSPs are divided into the following five different pathways depending on their social, economic and environmental development: SSP1 – sustainability, SSP2 – middle of the road, SSP3 – regional rivalry, SSP4 – inequality, SSP5 – fossil fuel development. An assumption about the degree of air pollution control (strong, medium or weak) is included on top of the baseline pathway, with stricter air pollution controls assumed to be tied to economic development (Rao et al., 2016). Weak air pollution controls occur in SSP3 and SSP4, with medium controls in SSP2 and strong air pollution controls in SSP1 and SSP5 (Gidden et al., 2019). A particular climate mitigation target, in terms of an anthropogenic radiative forcing by 2100, and the range of emission mitigation measures associated with achieving it are included in addition to the existing policy measures within each baseline SSP scenario. Climate mitigation targets vary from a weak-mitigation scenario with an anthropogenic radiative forcing of 8.5 Wm-2 by 2100, comparable with a 5 ∘C temperature change (Riahi et al., 2017), to a strong-mitigation scenario with a radiative forcing of 1.9 Wm-2 by 2100, in accordance with the Paris Agreement for keeping temperatures below 2 ∘C (United Nations, 2016). Some climate mitigation targets are comparable with those of the RCPs used in CMIP5 (2.6, 4.5 and 6.0), whilst others are new; e.g. ssp534-over is included as a delayed mitigation scenario. A scenario specific to the Aerosol and Chemistry Model Intercomparison Project (AerChemMIP), ssp370-lowNTCF, is also included to study the impact of mitigation measures to specifically control SLCFs on top of ssp370. Future biomass-burning emissions vary in each scenario, depending on the particular land-use assumptions (Rao et al., 2017). Whilst future anthropogenic and biomass-burning emissions are prescribed in each CMIP6 model from the same dataset, other natural emissions, e.g. dust, biogenic volatile organic compounds (BVOCs) etc., will be different and depend on the individual model configuration.

Figure 1 shows the future changes in global total (anthropogenic and biomass) emissions of the major air pollutant precursors across all of the CMIP6 scenarios, provided as input to the CMIP6 models. The overlying feature is that global air pollutant emissions are predicted to reduce across the majority of scenarios by 2100. The exception to this is that global and regional emissions increase or remain at present-day levels for ssp370 (Figs. 1 and 2). Some air pollutant emissions increase in the near term in other scenarios, e.g. nitrogen oxides (NOx) in ssp585 (by up to 15 %), but by 2100 these have been reduced. Future CH4 abundances show the largest diversity amongst the SSPs. Large increases in global CH4 abundances of more than 50 % are predicted for the fossil-fuel-dominated pathways of ssp370 and ssp585, whereas large reductions of ∼50 % are predicted to occur in the strong-mitigation scenarios of SSP1.

For SO2 large reductions of more than 50 % are shown for most scenarios and across most regions (Fig. 2), apart from Africa and Asia in ssp370. Near-term (2050) increases in SO2 occur over South Asia and other developing regions, which are then reduced in the latter half of the 21st century. Over Europe and North America, consistent decreases are predicted across all scenarios. The other major aerosol emissions, OC and BC, show similar reductions to SO2 across all scenarios and regions. For all aerosol and aerosol precursors, a reduction of 80 %–100 % (relative to 2015) in regional emissions is predicted by 2100 in the strong-mitigation scenarios. Changes in the emissions of the O3 precursors – NOx, CO and non-methane volatile organic compounds (NMVOCs) – show a similar increase across most regions for ssp370 but a general decrease in other scenarios. The change in these emissions is particularly diverse across all the scenarios in South Asia, with large relative increases in ssp370 (of up to 50 %) in contrast to the large decreases in ssp126 (up to 40 %). Across East Asia there is a 20 % increase in NOx emissions for ssp370 in 2050 but a long-term reduction across all scenarios.

Surface concentrations of O3 and PM2.5 have been obtained from all the CMIP6 models that made appropriate data available to the Earth System Grid Federation (ESGF) at the time of writing. To study changes in surface air pollutants over the industrial period, data have been obtained from the coupled historical simulations (Eyring et al., 2016) over the period 1850 to 2014 from all of the available ensemble members of each available CMIP6 model. For each model, a mean is taken using all available ensemble members prior to the calculation of the multi-model mean. For model evaluation purposes, 10 years of data from historical simulations have been used over the period that is relevant to the particular observational dataset (2000–2010 for ground-based PM2.5, 2004–2014 for PM2.5 reanalysis product and 2005–2014 for ground-based O3). To investigate future changes in air pollutants, all available data have been obtained over the period 2015 to 2100 for each of the different future coupled atmosphere–ocean model experiments, conducted as part of ScenarioMIP (O'Neill et al., 2016). CMIP6 model data have also been obtained for the AerChemMIP specific ssp370-lowNTCF scenario, which was only required to be conducted over the period 2015–2055 (Collins et al., 2017).

Concentrations of both pollutants at the surface have been obtained by extracting the lowest vertical level of the full 3D field output on the horizontal and vertical grid of each model (the “AERmon” CMIP6 table ID). For O3, this is supplied as a separate diagnostic which can be used directly. However, models contributing to CMIP6 will not all directly output PM2.5, and the calculation of PM2.5 will not be consistent across individual models due to the different treatment of aerosols and their components. For example only a few CMIP6 models include the simulation of ammonium nitrate in their aerosol scheme (currently, only GISS-E2-1-G and GFDL-ESM4 have provided nitrate mass mixing ratios on the ESGF database). Therefore, to use a consistent definition across all models, we calculated PM2.5 offline. In this study, surface PM2.5 is defined as the sum of the individual dry aerosol mass mixing ratios of black carbon (BC), total organic aerosol (OA – both primary and secondary sources), sulphate (SO4), sea salt (SS) and dust (DU) from the lowest model level extracted from the full 3D model fields. All BC, OA and SO4 aerosol mass is assumed to be present in the fine size fraction (<2.5µm), whereas a factor of 0.25 for SS and 0.1 for DU has been used to calculate the approximate contribution from these components to the fine aerosol size fraction (Eq. 1). 1PM2.5=BC+OA+SO4+0.25×SS+0.1×DU The factors used to calculate the contribution of SS and DU concentrations to the PM2.5 size fraction are likely to depend on the individual aerosol scheme and the simulated aerosol size distribution within a particular model. The calculation of an approximate PM2.5 concentration using Eq. () is therefore likely to introduce some errors, but it does provide an estimate that is consistent across models and also with that previously used in CMIP5 and ACCMIP (Fiore et al., 2012; Silva et al., 2013, 2017). For the CNRM-ESM2-1 model, anomalously large concentrations were obtained from the sea salt mass mixing ratios. Sensitivity tests with this model suggested that a much smaller factor of 0.01 was more appropriate to use for SS, which takes into account the non-dry nature of the sea salt aerosols and the large possible size range, up to 20 µm in diameter, of sea salt particles within the CNRM-ESM2-1 model (Pierre Nabat, personal communication, 27 November 2019).

Details of the data used in this study from different CMIP6 models, in both the historical and future scenarios, are presented in Table 1. For the historical period, data were available from 6 different CMIP6 models for O3 and 11 models for PM2.5. The future scenario with the most data available was ssp370, with 6 models supplying data for O3 and 10 models for PM2.5. For the other Tier 1 CMIP6 scenarios (ssp126, ssp245 and ssp585), data were only available for four models for O3 and seven for PM2.5 (all components). It was decided to focus the analysis on ssp370 and other Tier 1 scenarios due to the limited availability of model data for Tier 2 scenarios (ssp119, ssp434, ssp460 and ssp534-over). The results from an O3 parameterisation (Turnock et al., 2018, 2019), referred to in this study as HTAP_param, have also been included in the analysis of surface O3 from CMIP6 models for both the historical and future scenarios. The HTAP_param was previously developed based upon the source–receptor relationships of O3 derived from perturbation experiments of regional precursor emissions and global CH4 abundances (Wild et al., 2012; Turnock et al., 2018). The HTAP_param applies the fractional change in global CH4 abundance and regional-emission precursors (NOx, CO and NMVOCs) for a particular scenario to the ozone response from each individual model used in the parameterisation. The total O3 response is obtained by summing up the response from each of the individual models to all precursor changes across all source regions. The surface O3 response previously calculated from the HTAP_param in both the historical and future CMIP6 scenarios is compared to that from the CMIP6 models (Turnock et al., 2019). The O3 parameterisation does not take into account the effects of climate change on surface O3 concentrations and therefore provides an estimate of the emission-only-driven changes to surface O3, which we compare to the climate and Earth system models.

Present-day surface O3 and PM2.5 simulated by all of the CMIP6 models are evaluated against surface observations to ascertain model biases and inter-model discrepancies. Surface O3 observations are obtained from the database of the Tropospheric Ozone Assessment Report (TOAR; Schultz et al., 2017). The TOAR database provides a gridded product of surface O3 observations over the period 1970 to 2015. The majority of measurement sites are located in North America and Europe, with a smaller number of other sites in East Asia, Australia, New Zealand, South America, Southern Africa, Antarctica and remote ocean locations. Here we compile a monthly mean climatology of all available O3 observations over the period 2005–2014 from measurement locations that are classified as rural in the TOAR database (Schultz et al., 2017). The rural locations were selected to be representative of background (i.e. non-urban) O3 concentrations and are considered to be more appropriate in evaluating the simulated values obtained at the relatively coarse horizontal resolution of the global ESMs. Simulated surface O3 concentrations from the CMIP6 models are re-gridded onto the same resolution of the observational product (2∘×2∘) for evaluation purposes.

Surface PM2.5 observations have been obtained from all of the locations compiled in the database of the Global Aerosol Synthesis and Science Project (GASSP; http://gassp.org.uk/data/, last access: 2 July 2020, Reddington et al., 2017) to evaluate CMIP6 models. Background, non-urban PM2.5 data are compiled in the GASSP database from three major networks: the Interagency Monitoring of Protected Visual Environments (IMPROVE) network in North America, the European Monitoring and Evaluation Programme (EMEP), and Asia-Pacific Aerosol Database (A-PAD). Again, like for O3, the networks and observations for PM2.5 were selected to be representative of non-urban environments, which are more appropriate for the evaluation of global ESMs. With the exception of the IMPROVE network, most measurements of PM2.5 began after the year 2000. Like for O3, we compile a monthly mean climatology of PM2.5 but now over the period of 2000 to 2010, selected because the GASSP database contained the most observations within this period. Simulated surface PM2.5 was computed from CMIP6 models over the same time period as the observations and linearly interpolated to each measurement location. Whilst the surface observations measure total PM2.5 mass, the computed PM2.5 from CMIP6 models uses Eq. (1) and does not include all observable PM2.5 aerosol components (e.g. nitrate aerosol). Therefore, it is anticipated that the CMIP6 models will underrepresent the PM2.5 observations in this comparison.

To address the anticipated disparity between the observed ground-based PM2.5 and the approximate PM2.5 from CMIP6 models, a further comparison has been made between the CMIP6 models and the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), aerosol reanalysis product (Buchard et al., 2017; Randles et al., 2017). The MERRA-2 aerosol product assimilates observations of aerosol optical depth (AOD) from ground-based and satellite remote-sensing platforms into model simulations that use the GEOS-5 atmospheric model coupled to the GOCART aerosol module. The data assimilation used in MERRA-2 generally improves comparisons of PM2.5 with observations, but there are still overestimations due to dust and sea salt and underestimations over East Asia (Buchard et al., 2017; Provençal et al., 2017). Separate mass mixing ratios for BC, OA, SO4, SS and DU aerosol components are provided from MERRA-2, which are then combined using the formula in Eq. (1) to make an approximate PM2.5. Monthly mean approximate PM2.5 concentrations are then computed over the period 2005–2014 from the MERRA-2 reanalysis product to provide a more direct comparison and enhanced spatial coverage against the approximate PM2.5 concentrations calculated from the CMIP6 models calculated over the same time period.

The six CMIP6 models with data available for the historical experiments are evaluated against surface O3 observations from the TOAR database over the period 2005–2014. A long-term evaluation of surface O3 concentrations from CMIP6 models using observations compiled over the 20th century is presented separately in Griffiths et al. (2020). Figure 3 shows the annual and seasonal multi-model mean in surface O3 over the period 2005–2014 and the SD across the six CMIP6 models. The annual and seasonal mean surface O3 concentrations and evaluation against observations for individual CMIP6 models are shown in Figs. S2–S7. Higher surface O3 concentrations are simulated in the Northern Hemisphere summer (June, July, August – JJA) when O3 formation is enhanced by increased photolytic activity and levels of oxidants as well as larger biogenic emissions. The hemispheric difference in surface O3 is smaller in December, January and February (DJF) when O3 production is less in the Northern Hemisphere but higher in the Southern Hemisphere. However, model diversity is larger in DJF (Fig. 3e) due to individual models simulating different seasonal cycles of O3, particularly UKESM1-0-LL which has the most pronounced seasonal cycle of all 6 models (Fig. S2).

The multi-model mean of CMIP6 models overestimates surface O3 concentrations by up to 16 ppb annually and in both seasons when compared to observations from the TOAR database, although they do capture the broad hemispheric gradient in O3 concentrations (Fig. 3c, f and i). The model observational comparison of CMIP6 models to the TOAR observations is consistent across all models and with the previous evaluation of ACCMIP models (Young et al., 2018). This indicates a common source of error within models, for example uncertainties in emission inventories, deposition processes or vertical mixing (Wild et al., 2020). In addition, the coarse resolution of the ESMs could lead to an overproduction of O3 across polluted regions, with finer resolutions exhibiting improvements in the simulation of surface O3 (Wild and Prather, 2006; Neal et al., 2017). Smaller model biases exist in DJF (<5 ppb) than in JJA (5–15 ppb), mostly attributed to the strong seasonal cycle simulated by UKESM1-0-LL. In contrast to other models (Figs. S2–S7), UKESM1-0-LL underpredicts surface O3 in DJF over most continental Northern Hemisphere locations, potentially indicating there is excessive NOx titration of O3 in this model, which is also shown by the large sensitivity of O3 formation to NOx concentrations over the historical period (Fig. S17).

The observed annual cycle in surface O3 averaged across measurement locations within different regions is compared to that simulated by CMIP6 models (Fig. 4). Across most regions, the mean annual cycle from CMIP6 models compares relatively well to that observed. The overprediction of surface O3 values in JJA is evident across most regions, as are the large concentrations in BCC-ESM1 and GISS-E2-1-G and the strong seasonal cycle in UKESM1-0-LL across Northern Hemisphere continental regions. Additionally, the timing of peak O3 over continental Northern Hemisphere locations occurs earlier in the observations (springtime) than in the CMIP6 models (spring and summer), which is consistent with that from ACCMIP models (Young et al., 2018). At oceanic observation locations, surface O3 is overestimated in CMIP6 models by up to 20 ppb across all seasons, indicating that the O3 deposition rate could be underestimated here (Clifton et al., 2020). There is also a large overestimation (∼20 ppb) in all models at the one observation location in South East Asia, potentially due to difficulty in simulating O3 in the maritime continental boundary layer using lower-resolution global ESMs. In contrast to this, CMIP6 models, particularly UKESM1-0-LL and GISS-E2-1-G, tend to underpredict the observed surface O3 concentrations at locations in the South Pole region in JJA by ∼5 ppb. This could be due to lack of long-range transport of O3 to these sites, inaccuracies in Southern Hemisphere precursor emissions or because of the difficulty in simulating O3 concentrations at the appropriate elevation of measurement sites located on the Antarctic Ice Sheet.

The simulated changes in surface O3 across six CMIP6 models and the HTAP_param are shown in Fig. 9 and Figs. S14–S15 over the historical period of 1850 to 2014. The CMIP6 multi-model mean shows that global annual mean surface O3 has increased by 11.7±2.3 ppb since 1850 (±1 SD), although the change could be as large as 14 ppb (from BCC-ESM1) or as little as 7 ppb (from UKESM1-0-LL). Globally and over most regions there has been a larger historical increase in surface O3 in JJA than in DJF (Fig. S16). The 1850 to 2000 multi-model annual mean change in surface O3 from the CMIP6 models of 10.6 ppb is in good agreement with the 10±1.6 ppb simulated by the CMIP5 models used in ACCMIP (Young et al., 2013). An evaluation of the long-term changes in surface O3 over the historical period simulated by the CMIP6 models at specific measurement locations is presented separately in the tropospheric O3 CMIP6 companion paper of Griffiths et al. (2020). This shows that CMIP6 models can reasonably represent long-term changes in surface ozone since the 1960s, providing a degree of confidence in the future projections of changes in the CMIP6 scenarios. However, long-term changes in simulated surface O3 from the previous generation of global coupled chemistry–climate models (used in CMIP5) were found to underestimate the observed trend at Northern Hemisphere monitoring locations (Parrish et al., 2014). Further comparisons of historical surface O3 simulated by CMIP6 models with long-term historical observations are outside the scope of the current work but will be the subject of future research.

A large diversity in the simulated historical changes is shown across the different regions analysed here, with UKESM1-0-LL tending to simulate the smallest historical change and GISS-E2-1-G or BCC-ESM1 the largest. The large diversity across CMIP6 models in the surface O3 response over the historical period can be attributed to the different magnitude of simulated O3 concentrations in the 1850 period (Fig. S14) and the rate of change in regional mean O3 concentrations (Fig. S15), which is related to the different chemical sensitivity of O3 formation in each model to changing NOx concentrations (Fig. S17). Larger differences between CMIP6 models are shown in the DJF mean historical changes over Northern Hemisphere regions than what occurred in JJA (Fig. S16), reflecting the differences shown in the model evaluation (Fig. 4) and the strong seasonality of the changes. Even though the historical surface O3 response is small in UKESM1-0-LL, it is shown to have larger tropospheric changes in O3 over the historical period compared to other CMIP6 models (Griffiths et al., 2020).

South Asia is the region with the largest diversity in simulated historical changes in surface O3, between 16 and 40 ppb, with a larger range in DJF (10–40 ppb) than in JJA (19–36 ppb). The large diversity in CMIP6 models is attributed to the large differences in simulated NOx concentrations and hence chemical sensitivities of O3 formation occurring across South Asia over the historical period (Fig. S17). In addition, the large historical change in PM2.5 over this region (Fig. S18) could alter the heterogeneous loss rate of radicals to aerosols and therefore also affect O3 formation. Surface O3 is simulated to have increased by between 10 and 30 ppb on an annual mean basis and by a larger amount in JJA (12 to 37 ppb) over the major northern anthropogenic source regions since 1850, driven mainly by the large increases in anthropogenic precursor emissions of CH4, NOx, CO, and NMVOCs over this period.

A qualitative estimate of the influence of non-emission-driven processes (chemistry and climate change) can be ascertained by comparing results from the HTAP_param, an emission-only-driven model, to those of the CMIP6 models. Simulated historical changes in surface O3 from UKESM1-0-LL are comparable to those from the HTAP_param, indicating that the magnitude of change simulated by UKESM1-0-LL is similar to that solely from changes in precursor emissions. However, the global annual mean surface O3 response of 7.6±0.7 ppb from HTAP_param over the historical period is 4.1 ppb lower than the CMIP6 multi-model mean, indicating globally that non-emission-driven processes have contributed to approximately 30 % of the change in surface O3, although this contribution varies regionally. The different magnitude of response across models could be due to non-emission-driven processes, e.g. from different chemistry schemes and climate change signals within models.

The simulated change in annual mean surface PM2.5 across 11 CMIP6 models is shown in Fig. 10 over the historical period of 1850 to 2014. CMIP6 models simulated an increase in global annual and seasonal mean surface PM2.5 concentrations of <2 µgm-3 (15 %–20 %) since 1850. Larger regional increases in surface annual mean PM2.5 of up to 12 µgm-3 are simulated across South and East Asia, with changes in DJF (up to 21 µgm-3) larger than those in JJA (up to 12 µgm-3; Fig. S16), reflecting the strong seasonality of PM2.5 concentrations in these regions. The historical increase in surface PM2.5 is primarily driven by the large increase in anthropogenic aerosol and aerosol precursor emissions over the 1850–2014 period (Hoesly et al., 2018). The largest model diversity is also exhibited over the Asian regions, with variations in the response between models of up to 50 % and larger differences between models in DJF than JJA (Fig. S16), reflecting the differences shown in the present-day model evaluation (Fig. 6). The inter-model differences can be attributed to the different simulation of historical changes in the anthropogenic components sulphate, black carbon and organic aerosols (Fig. S18). The largest inter-annual variability in surface PM2.5 concentrations occurs over the North African and Middle East regions as they are located near large sources of dust, whose emissions are highly dependent on meteorological fluctuations (wind speed). Over Europe and to a lesser extent Russia, Belarus, Ukraine and North America, the increase in surface PM2.5 concentrations since 1850 peaked in the 1980s at 4 µgm-3 above the 2005–2014 mean value before decreasing over the last 30 years. There are limited long-term multi-decadal observational data available to assess changes in aerosols simulated by global models. Previous studies using long-term data since the 1980s, mainly over Europe and North America, have found that global models are able to reproduce the observed multi-decadal changes in aerosols relatively well (Pozzoli et al., 2011; Leibensperger et al., 2012; Tørseth et al., 2012; Chin et al., 2014; Turnock et al., 2015; Aas et al., 2019). More recently, global composition models, including some CMIP6 models, were shown to be able to reproduce the observed changes in AOD, sulphate and particulate matter over the last 2 decades (Mortier et al., 2020). The ability of global composition models to reproduce historical changes in aerosols provides a degree of confidence in the future projections under the CMIP6 scenarios. Further model observational comparisons of multi-decadal changes in aerosols will need to be undertaken to improve the understanding of changing aerosol properties and processes.

Global annual mean surface O3 is reduced by more than 5±1.2 ppb (±1 SD value of the multi-model mean) in the near-term (2050) and by 9±1.6 ppb in 2100 in the strong air pollutant and climate mitigation scenario ssp126 (Fig. 11). Smaller reductions in global annual mean surface O3 of 4±1.7 ppb are predicted for the middle-of-the-road pathway (ssp245) by 2100, whereas for the weak climate and air pollutant mitigation scenario ssp370, a global annual mean increase in surface O3 of 1.6±0.9 ppb in 2050 and 0.6±1.0 ppb by 2100 is predicted. However, implementing strong emission controls for SLCFs on top of a weak climate mitigation scenario (ssp370-lowNTCF) shows that previous increases in global annual mean surface O3 can be substantially reduced to values that are 2.5±0.5 ppb below the 2005–2014 mean value in 2050, with benefits to air quality and climate (Allen et al., 2020). For ssp585, which has weak climate mitigation measures but strong air pollution controls, a near-term increase in global annual mean surface O3 of 1.4±0.8 ppb is predicted in 2050, but by 2100 surface O3 reduces by 2.7±1.5 ppb, relative to 2005–2014, due to the implementation of air pollutant controls in the latter half of the 21st century.

The global response in annual mean surface O3 concentrations to the different scenarios is also repeated across the different world regions, albeit with differing magnitudes. In ssp370, increases in annual mean surface O3 are predicted to occur across North America (+1.6 ppb), Europe (+5.4 ppb) and East Asia (+5.9 ppb), with the largest increase, of 15.1±9.6 ppb, predicted in South Asia by 2100. Despite the reductions in O3 precursor emissions across North America, Europe and East Asia by 2100 (Fig. 2), surface O3 concentrations have continued to increase up to the end of this period, indicating the importance of future changes in chemistry, global CH4 abundances and climate to the response of surface O3 in ssp370 (Wild et al., 2012; Gao et al., 2013; Rasmussen et al., 2013; Young et al., 2013; Colette et al., 2015; Fortems-Cheiney et al., 2017; Li et al., 2019; Turnock et al., 2019). South Asia shows the largest increase in surface O3 as precursor emissions are anticipated to increase across this region on top of the large climate change signal and growth in CH4 abundance. Additionally, the largest diversity in projections between the CMIP6 models is shown over South Asia, indicating that there is some disagreement between the models as to the magnitude and extent of changes over this region. Surface O3 across oceanic regions (background) is predicted to remain at or near current values in ssp370 due to the increases in water vapour in a warming world, leading to more O3 destruction (Johnson et al., 1999; Doherty et al., 2013). The impact of more aggressive near-term reductions to emissions of SLCFs (but not CH4) on top of the ssp370 pathway is shown by the smaller changes in the ssp370-lowNTCF (Figs. 11 and S19–S20 for individual models). In this pathway, surface O3 concentrations are reduced globally and across most regions to be at or near 2005–2014 values, a substantial benefit to surface O3 air quality compared to ssp370. Surface O3 concentrations are predicted to have almost halved by 2050 across South Asia in ssp370-lowNTCF. However, across East Asia the additional precursor emission reductions in ssp370-lowNTCF have resulted in smaller benefits to surface O3 concentrations being simulated by the CMIP6 models than in other regions (Fig. S20), which is attributed to an increase in surface O3 concentrations over eastern China (a part of the larger East Asian region shown in Fig. S1). This increase in surface O3 results from the slight increase in NMVOC emissions (Fig. 2) and a reduction in the NOx titration of O3 due to the large decreases in NOx emissions in ssp370-lowNTCF. In addition, a reduction in the heterogeneous loss of radicals due to decreases in PM2.5 concentrations in ssp370-lowNTCF could also lead to increased surface O3 concentrations (Li et al., 2019).

Surface O3 concentrations predicted across Northern Hemisphere regions in ssp585 are similar to ssp370 due to comparable changes in air pollutant emissions and climate change. However, a notable exception is a reduction in surface O3 across regions towards the latter half of the 21st century (post-2080), when there are additional reductions in precursor emissions and global CH4 abundances by 2100. Surface O3 shows a slightly slower increase until the mid-21st century over South Asia in ssp585 than what occurred in ssp370. This can be attributed to a slightly different temporal evolution of NOx emissions over this region in that they peak earlier (by 2040) and decline more rapidly in ssp585 when compared to the continual increase in NOx emissions in ssp370 (Fig. 2), which results in a different response of O3 formation within CMIP6 models, In addition, there are more CMIP6 models with data available for ssp370 (six models) than ssp585 (four models; Table 1), which could affect the multi-model mean response shown in Fig. 11.

The future scenario ssp245 (middle of the road) predicts annual mean surface O3 concentrations that tend to remain at or near the 2005–2014 mean values by 2100 across the major anthropogenic source regions of the Northern Hemisphere, whereas for other tropical and Southern Hemisphere regions surface O3 concentrations are reduced by more than 4 ppb. The changes in ssp245 are driven by larger precursor emission controls, a smaller climate change signal and controlling CH4 so that global abundances are below 2015 values by 2100 (Fig. 1g). In ssp245 a near-term (up to 2040) increase in surface O3 is shown across East Asia and South Asia, which could be attributed to the peaking of global CH4 abundances at this point prior to then reducing.

The Tier 1 future scenario with the strongest climate and air pollutant mitigation measures, ssp126, shows substantial decreases in surface O3 concentrations across most regions due to the large reduction in precursor emissions, global CH4 abundances and small climate change signal. Reductions in surface O3 of more than 10 ppb are predicted across anthropogenic-emission source regions of the Northern Hemisphere, with smaller reductions across Southern Hemisphere regions.

Projections from the CMIP6 models show that to achieve global benefits for regional surface O3, it is important to control O3 precursor emissions (including CH4) in addition to limiting future climate change. However, scenarios with large climate change signals (ssp370 and ssp585) but different post-2050 controls on O3 precursors (most notably CH4 and NOx) show different long-term changes in regional surface O3 concentrations, which could have important consequences for any potential human health impacts.

A more detailed comparison of future surface O3 projections between CMIP6 models has been undertaken for ssp370 as it is the scenario with the largest number of available models (Table 1). The regional change in annual and seasonal mean surface O3, relative to 2005–2014, in 2050 (2045–2055 mean) and 2095 (2090–2100 mean) for ssp370 from four CMIP6 models and the HTAP_param is shown in Fig. 12. An analysis of the relationships, in terms of correlation coefficients, between future annual mean surface O3 concentrations and other variables (CH4 concentrations, surface air temperature, NOx concentrations, emissions of BVOCs and anthropogenic emissions of NMVOCs) is undertaken for CMIP6 models in the ssp370 scenario (Fig. 13). Discrepancies in the simulated response of background O3 across the ocean region (as well as the South Pole and Pacific, Australia and New Zealand) are noticeable between individual models, with UKESM1-0-LL predicting a decrease in surface O3 compared to the small increase from the HTAP_param and most other models in both 2050 and 2095 (Fig. S19). The future surface O3 response in UKESM1-0-LL over the ocean region exhibits a large negative correlation with surface temperature changes (Fig. 13), indicating the importance of future climate change in this model over remote regions. UKESM1-0-LL is a model with high equilibrium climate sensitivity (ECS; 5.4 K) compared to other CMIP6 models (Forster et al., 2019; Sellar et al., 2019) and therefore will exhibit a larger climate response (surface temperature and water vapour), leading to enhanced background O3 destruction via water vapour and the hydroxyl radical (OH). Over the North Pole region all models show surface O3 increases that are larger than the HTAP_param, with a larger increase in DJF than JJA. The large future temperature response over the Arctic and changes in NOx concentrations and emissions of NMVOCs are particularly important drivers of surface O3 changes across most CMIP6 models in this region with comparatively low local emissions (Fig. 13).

Differences in the predicted surface O3 between models exist across South Asia, where CESM2-WACCM (and BCC-ESM1 in 2050) predicts a response that is twice as large as UKESM1-0-LL and GFDL-ESM4. The lower annual mean response over South Asia in UKESM1-0-LL and GFDL-ESM4 is driven by a reduction in DJF in these models (Fig. S21), which results in the DJF change in 2050 being lower than the 2005–2014 annual mean value (Fig. 12). The large increase in NOx emissions in ssp370 over South Asia (∼80%) has resulted in areas of NOx titration, particularly in DJF, near the Indo-Gangetic Plain in both UKESM1-0-LL and GFDL-ESM4, reducing surface O3 concentrations (Figs. S19 and S21). This strong feature of NOx titration of O3 in DJF is absent in both CESM2-WACCM and BCC-ESM1, resulting in larger O3 production over South Asia. The comparison in Fig. 12 shows how the O3 chemistry within models responds differently across a particular area in a future scenario with a large climate change signal and over a region with large increases in local precursor emissions but that all the drivers related to regional O3 change in South Asia are similarly important across all models (Fig. 13).

Over South America and Southern Africa, particularly the tropical areas (Fig. S19), larger future changes in surface O3, particularly by 2100, are predicted by GFDL-ESM4 and UKESM1-0-LL than by CESM2-WACCM. These changes over South America are larger in JJA in all models, with small seasonal differences over Southern Africa. Over this region, biogenic emissions (particularly isoprene) are an important source of O3 formation. Discrepancies in the future response of these BVOC emissions between models could be occurring due to the differing magnitudes of climate and land-use change and how they are coupled within individual CMIP6 models (Table S1), which could affect future surface O3. Future changes in the total emissions of BVOCs and those solely from isoprene obtained from five CMIP6 models (Figs. S22 and S23) show that CESM2-WACCM has larger total BVOC emissions over the period 2005–2014 (due to the inclusion of more BVOCs), which then increase in the future ssp370 scenario, along with isoprene emissions, resulting in a smaller increase (and even decrease in some parts of the region) in O3. In contrast, UKESM1-0-LL shows a larger increase in O3 and a reduction in BVOC emissions, mainly from isoprene (Fig. 23), over parts of South America and tropical Africa. Figure 13 shows that there are differing relationships between future surface O3 concentrations, BVOC emissions and NOx concentrations across CMIP6 models over South America and Southern Africa. Over Southern Africa, UKESM1-0-LL shows a different relationship between BVOC emissions and surface O3 concentrations than other CMIP6 models, indicating that this could be leading to the different future O3 response in this model over this region. Similarly, Fig. 13 shows that over South America, CESM2-WACCM has a different relationship between surface O3 and the variables considered here than in other CMIP6 models, particularly for BVOCs, leading to the different future responses in this model over this region. Figure 13 shows that there are differences between models in the surface O3 response over regions such as South America and Southern Africa, which are potentially linked to the land-surface response and are important to understand more in future work.

Whilst there are disagreements between models over some regions, there is also substantial consistency in the predicted increase in annual mean surface O3 in ssp370 over North America, Europe and East Asia, which is larger than that from HTAP_param. However, BCC-ESM1 tends to predict a larger increase than the other three models, potentially due to the coarser resolution of this ESM. There are differences in simulated seasonal response across these regions, with all models showing a smaller increase in JJA than DJF across North America and Europe, whilst across East Asia there tends to a be a larger future surface O3 increase in JJA than DJF. Figure 13 shows that there is a negative correlation between surface O3 and NOx concentrations as well as between O3 and NMVOC emissions for most CMIP6 models across these regions, reflecting that, as most anthropogenic precursor emissions (including NOx) decrease in this scenario (Fig. 2), surface O3 is simulated to increase. An exception to this is across East Asia, where the increase in NMVOC emissions in ssp370 (Fig. 2) is positively correlated with surface O3, indicating different chemical drivers of future O3 across this region. In addition, there are positive correlations between the other variables (temperature, CH4 and BVOCs) for most CMIP6 models, indicating that changes in climate and global CH4 abundances are also important drivers of surface O3 increases over these regions.

The differences between the individual CMIP6 models highlight the importance of further understanding how future O3 chemistry is affected by changes to precursor emissions and climate. The predicted differences in models can be quite pronounced over regions like South Asia, where changes in one model can be double that of another model, which could have important consequences for the ability of models to simulate future regional air quality.

Relatively small global changes in annual mean surface PM2.5 are predicted for all CMIP6 models across all scenarios (Fig. 14), with an increase in ssp370 and a reduction in the others. Small reductions in PM2.5 are predicted for all scenarios across Europe (0.3 to 3 µgm-3) and North America (0.0 to 1.3 µgm-3) due to the reduction in aerosol and aerosol precursor emissions. Differences in PM2.5 between scenarios are highlighted across a number of regions.

For the weak climate and air pollutant mitigation scenario ssp370, increases in annual mean surface PM2.5 are predicted across South Asia (7.3±4.1 µgm-3 by 2050 and 3.1±3.1 µgm-3 by 2100), South East Asia (2.7±4.7 µgm-3 by 2100), Southern Africa (1.6±3.7 µgm-3 by 2100), Central America (2.8±3.2 µgm-3 by 2100) and South America (2.9±3.6 µgm-3 by 2100). The increases in PM2.5 are driven mainly by the increase in aerosol and aerosol precursor emissions in this scenario (Fig. 2), shown by the positive correlations between emissions and surface PM2.5 in CMIP6 models across these regions (Fig. 16). However, there is a degree of uncertainty associated with all of these future projections, indicated by the large diversity across the CMIP6 models. Some of the largest predicted increases in surface PM2.5 occur across South Asia in ssp370, a region with already high present-day PM2.5 concentrations. The increase in PM2.5 peaks in 2050 across this region, which coincides with the increase in SO2, BC and OC emissions before declining to 2100, when emissions reduce. Over East Asia, annual mean PM2.5 concentrations are simulated to remain at or near 2005–2014 values until the latter half of the 21st century, when the decrease in emissions reduces PM2.5 concentrations by 2.5±2.7 µgm-3. The impact of reductions in SLCFs on top of the ssp370 scenario acts to constrain any increases in PM2.5 concentrations to near present-day values across most regions. However, substantial reductions in PM2.5 concentrations of 5.6±2.0 µgm-3 and 5.3±2.1 µgm-3 below 2005–2014 values are achieved by 2050 across East and South Asia, respectively, by implementing these measures. Due to the short lifetime of aerosols in the atmosphere, PM2.5 concentrations respond rapidly to the large cuts in emissions that occur in ssp370-lowNTCF and show the benefits of targeting these emissions, although there could be a potential climate impact (Allen et al., 2020).

Reductions in annual mean surface PM2.5 are simulated across all regions for ssp126, ssp245 and ssp585. Differences exist in the magnitude and timing of PM2.5 reductions across regions linked to the changes in emissions. The largest reductions in PM2.5 occur over South Asia in 2100 and range from 11.1±2.8 µgm-3 in ssp126 to 8.6±2.9 µgm-3 in ssp585, a substantial benefit to regional air quality. Similar benefits to PM2.5 are achieved over East Asia by 2100, although the more rapid improvements occur over this region in the first part of the 21st century.

The response of PM2.5 concentrations is more variable, with a larger diversity across CMIP6 models within regions that are close to natural aerosol emission sources. This is particularly noticeable over North Africa, where the variability across CMIP6 models in dust emissions from the Saharan source region (Fig. S8) results in an uncertain PM2.5 response across this region. A similar response is also exhibited across the Middle East and Central Asia. The potential influence of BVOCs on SOA formation (Figs. S22 and S26) could also be contributing to the diversity in the CMIP6 model responses across the South America and Southern Africa regions.

The CMIP6 models show that future reductions in aerosols and aerosol precursors will lead to a decrease in surface PM2.5 concentrations across most world regions and a benefit to regional air quality (and human health), consistent with that from CMIP5. However, if emissions are not controlled over economically developing regions such as South America, Asia and Africa, then surface PM2.5 is anticipated to increase and worsen future regional air quality. Targeting emission reductions in SLCFs in the short term shows the potential for rapid improvements in surface PM2.5 and air quality.

In a similar analysis to that for surface O3, a more detailed comparison has been undertaken of four CMIP6 models predicting changes in annual and seasonal surface PM2.5 in 2050 and 2095 under ssp370 (Fig. 15). In addition, an analysis of the relationships, in terms of correlation coefficients, between future annual mean surface PM2.5 and other variables (total surface precipitation; surface air temperature; and emissions of BVOCs, SO2, BC and organic aerosol) has been undertaken for CMIP6 models in the ssp370 scenario (Fig. 16). Small reductions in annual mean surface PM2.5 concentrations (<2µgm-3) are simulated consistently by all CMIP6 models across North America and Europe in ssp370, with larger reductions simulated in DJF than JJA. The reductions in annual mean PM2.5 over Europe and North America are mainly attributed to decreases in the BC and SO4 components (Figs. S24 and S25), as indicated by the strong correlations with BC and SO2 emissions across CMIP6 models (Fig. 16). However, by 2095 a small increase (up to 2 µgm-3) is simulated in JJA by UKESM1-0-LL and CESM2-WACCM over North America, which could be attributed to changes in climate due to the strong positive correlations in both models for temperature, precipitation and BVOCs (Fig. 16).

South Asia, the region with the largest simulated future change in annual mean surface PM2.5, of up to 12 µgm-3, shows fairly good agreement between three CMIP6 models (UKESM1-0-LL, GFDL-ESM4 and CESM2-WACCM) as projections in 2050 and 2095 are all within the range of each of the individual models. The future increases in annual mean surface PM2.5 appear to be strongly driven by emission changes as there are strong positive correlations between these variables across South Asia in all models (Fig. 16). Across South Asia, all models simulate a larger increase in DJF mean surface PM2.5 concentrations, of up to 18 µgm-3 by 2050, than what occurs in JJA, reflecting the seasonality shown in the model evaluation. The MIROC-ES2L model predicts smaller future increases in surface PM2.5, of up to 5 µgm-3, than the other models across South Asia in both 2050 and 2095. This is a result of smaller changes in the BC, OA and sulphate aerosol components in the MIROC-ES2L model despite increases in aerosols and aerosol precursor emissions across South Asia in ssp370 (Fig. S24–S26).

Disagreements in both the sign and magnitude of simulated future annual and seasonal mean surface PM2.5 changes between CMIP6 models are also exhibited across East Asia. Small regional annual mean increases are predicted in 2050 due to PM2.5 increases in JJA from all models apart from GFDL-ESM4. A larger reduction in the SO4 component is simulated over East Asia by GFDL-ESM4 than in other models (Fig. S25), resulting in an overall decrease in PM2.5. In 2095 most models simulate a reduction in PM2.5 concentrations in both seasons across East Asia, apart from CESM2-WACCM due to the increase in JJA. All models simulate continual reductions out to 2100 for SO4 across this region, whereas BC increases in the near term before decreasing out to 2100. For OA, CESM2-WACCM shows larger increases over East Asia in both 2050 and 2095 compared to the other models, which show a smaller increase in 2050 and a reduction by 2095 (Fig. S26). CESM2-WACCM includes a more complex treatment of SOA formation, showing a strong response to climate and historical trends in OA (Tilmes et al., 2019). Positive correlations are shown for CESM2-WACCM between surface PM2.5 and emissions of BVOC as well as between surface PM2.5 and temperature (Fig. 16), which are not present in other models and could explain the differences between this model and others across East Asia. The discrepancies in CMIP6 models are not as obvious over South Asia as the effect of the increase in OA over South Asia in CESM2-WACCM is masked by coincident increases in other components across other models, as indicated by the strong correlations with emissions here. CESM2-WACCM also shows larger simulated increases in PM2.5 over South America, Central America, Southern Africa and South East Asia than other models, which can be attributed to the larger increase in the OA fraction (Fig. S26) and the strong correlations in this model with changes in temperature and emissions (BVOCs and SO2). Over Southern Africa UKESM1-0-LL shows a reduction in future PM2.5 in contrast to other models due to a reduction in the BC, OA and dust aerosol components (Figs. S24, S26 and S27). UKESM1-0-LL exhibits particularly strong negative correlations for surface PM2.5 when compared with temperature and precipitation. These relationships over Southern Africa are quite different to other CMIP6 models, which is also highlighted in the model evaluation over this region (Fig. 8) and indicates that climate change influences aerosol concentrations differently over this region in this model (Fig. 16). In addition, there is a slight positive correlation of PM2.5 with BVOC emissions in UKESM1-0-LL over Southern Africa. Future biogenic emissions (including monoterpenes) reduce here in ssp370 (Fig. S22), potentially due to land-use vegetation change as UKESM1-0-LL has dynamic vegetation coupled to BVOC emissions (Table S1 in the Supplement). This could also reduce PM2.5 concentrations over this region because monoterpene emissions are the main precursor to SOA formation in UKESM1-0-LL (Mulcahy et al., 2020).

The annual and seasonal mean PM2.5 response is variable across individual CMIP6 models over regions close to natural sources of particulate matter (North Africa; Central Asia; and Pacific, Australia and New Zealand). Over these regions there is a large range in both the sign and magnitude of the annual and seasonal PM2.5 response, which can be mainly attributed to the dust fraction (Fig. S27) and the fact that this aerosol source has a large inter-annual variability in its emission strength. There is also a lack of consistency across CMIP6 models in the correlations of PM2.5 with any individual driver, indicating the variability of aerosol sources in these regions within models. Interestingly, the CMIP6 models do not agree in the sign and magnitude of future changes to dust concentrations in ssp370 (Fig. S27).

Across the ocean and North Pole regions, all the CMIP6 models tend to simulate a small increase in PM2.5 concentrations, which can be attributed to increases in sea salt concentrations (Fig. S28). A strong increase in sea salt concentrations is simulated in all models across the Southern Ocean (and other oceans), potentially driven by changes to meteorological conditions (reflected by the positive correlations of PM2.5 with the climate variables temperature and precipitation in Fig. 16), which increase wind speed and sea salt emissions. As ssp370 is a scenario with a large climate change signal, the increases in PM2.5 across the North Pole, particularly in 2100, can be attributed to the melting of sea ice increasing sea salt emissions, which again is reflected in the positive correlations of PM2.5 with climate variables over this region. However, the magnitude of this response is different in the CMIP6 models due to the underlying ECS and the response of Arctic surface temperatures within the individual model.

The differences in the simulated future PM2.5 changes across the CMIP6 models in ssp370 highlight that it is important to consider how natural sources of aerosol respond in a future climate in addition to changes from anthropogenic emissions. Particular differences between models have been shown for dust, sea salt and also organic (secondary) aerosols, which should be explored further. In addition, the different representations of aerosols within individual models, e.g. organic aerosols, are an important consideration as they can make a large difference to any future regional projection of PM2.5.