The global impact of the transport sectors on atmospheric aerosol in 2030 – Part 1 : Land transport and shipping

Using the EMAC (ECHAM/MESSy Atmospheric Chemistry) global climate-chemistry model coupled to the aerosol module MADE (Modal Aerosol Dynamics model for Europe, adapted for global applications), we simulate the impact of land transport and shipping emissions on global atmospheric aerosol and climate in 2030. Future emissions of short-lived gas and aerosol species follow the four Representative Concentration Pathways (RCPs) designed in support of the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. We compare the resulting 2030 land-transportand shipping-induced aerosol concentrations to the ones obtained for the year 2000 in a previous study with the same model configuration. The simulations suggest that black carbon and aerosol nitrate are the most relevant pollutants from land transport in 2000 and 2030 and their impacts are characterized by very strong regional variations during this time period. Europe and North America experience a decrease in the land-transport-induced particle pollution, although in these regions this sector remains a major source of surface-level pollution in 2030 under all RCPs. In Southeast Asia, however, a significant increase is simulated, but in this region the surface-level pollution is still controlled by other sources than land transport. Shipping-induced air pollution is mostly due to aerosol sulfate and nitrate, which show opposite trends towards 2030. Sulfate is strongly reduced as a consequence of sulfur reduction policies in ship fuels in force since 2010, while nitrate tends to increase due to the excess of ammonia following the reduction in ammonium sulfate. The aerosol-induced climate impact of both sectors is dominated by aerosol-cloud effects and is projected to decrease between 2000 and 2030, nevertheless still contributing a significant radiative forcing to Earth’s radiation budget.


Introduction
The transport sectors, including land transport, shipping and aviation, are major sources of atmospheric pollution (e.g., Righi et al., 2013). The emissions from transport are growing more rapidly than those from the other anthropogenic activities. According to the ATTICA assessment (Uherek et al., 2010;Eyring et al., 2010), land transport and shipping shared ✿✿ 12 % of the global CO 2 emissions from transport in the year 2000, respectively. In the time period 1990-2007, the EU-15 CO 2 -equivalent emissions from land transport and shipping increased by 24 and 63 %, respectively. This growth is expected to continue in the future, due to increasing world population, economic activities and related mobility. The future road traffic scenarios analysed by Uherek et al. (2010) essentially agree in projecting an increase of both fuel demand and CO 2 emissions until 2030, with up to a factor of ∼ 3 increase in CO 2 emissions with respect to 2000. The ATTICA assesment also showed that emissions of CO 2 from land transport and shipping affects ✿✿✿✿✿✿ affect the global climate by exerting a radiative forcing (RF) effect of 171 (year 2000) and 37 mW m −2 (year 2005), respectively. These two sectors together account for 13 % of the total anthropogenic CO 2 warming (year 2005).
In addition to long-lived greenhouse gases, ground-based vehicles and ocean-going ships emit aerosol particles as well as a wide range of short-lived gases, including also aerosol precursor species. Atmospheric aerosol particles have significant impacts on climate, through their interaction with solar radiation and their ability to modify cloud microphyisical ✿✿✿✿✿✿✿✿✿✿✿✿ microphysical ✿ and optical properties (Forster et al., 2007). In populated areas, they also affect air quality and human health (Pope and Dockery, 2006;Chow et al., 2006).
The present work represents a follow-up study of Righi et al. (2013, hereafter R13). In that study, we considered year 2000 emissions and performed several sets of model simulations (i) to estimate transport impacts on atmospheric aerosol, (ii) to quantify the uncertainty in the effects on particle number concentrations related to the assumed particle size distribution of emitted particles, (iii) to explore non-linearities in the aerosol response to the perturbations induced by transport emissions; and (iv) to quantify the radiative forcings of transport-induced aerosol. These model experiments revealed that land transport is the most significant source of large-scale particulate matter pollution (mostly black carbon and nitrate) in Europe, the USA and the Arabian Peninsula, while it is less relevant in southern and eastern Asia, where other sources dominate. The impact of the shipping sector was found to be of high importance over the oceans in the Northern Hemisphere, especially for Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | sulfate and nitrate aerosol, with significant effects also along the coastlines. The transportinduced changes of aerosol number concentration turned out to be very sensitive to the assumed size distribution of emitted particles, with the largest uncertainty (spanning about one order of magnitude) simulated for the land transport sector. We found that the aerosol response to the emission perturbations is approximately linear for black carbon, particulate organic matter and particle number. In contrast, non-linearities were found for aerosol nitrate, sulfate and ammonium, due to the non-linear behaviour of the chemical processes involved in the formation of these species.
The aerosol impacts on the Earth's radiation budget induced by the transport sources were estimated in R13, in terms of all-sky and clear-sky global RF. For each sector, a range of possible RF values was calculated, according to the different assumptions on the size distributions of emitted particles. For land-transport and shipping, we simulated ranges of −80 to −12 mW m −2 and of −222 to −153 mW m −2 , respectively, with a dominant contribution from cloud effects in both cases.
The present study complements R13 by presenting the impacts of land-transport and shipping for different future scenarios. The impact of aviation emissions will be the subject of a companion study. In particular, we focus here on the year 2030 ✿✿✿✿✿✿✿✿✿✿ emissions based on the four Representative Concentration Pathways (RCPs, Moss et al., 2010;van Vuuren et al., 2011a). These scenarios were developed in support of the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR5). We choose the year 2030 in order to reduce the influence of the uncertainties in the emission data, as such uncertainties are known to increase with the projection time. In analogy to R13, we focus only on the effects of the emission of aerosol and aerosol precursors, while we refrain from considering the impact of climate change on aerosol distributions. This enables us to quantify the pure effect of emission changes on aerosol when comparing the 2030 simulations with the year 2000 experiments by R13. The model therefore is driven by prescribed concentrations of long-lived species and radiatively active gases representative of year 2000 conditions, and by meteorological data for the period 1996-2005 to nudge the dynamics. Although , ✿✿✿✿✿✿✿✿✿ although ✿ the impact of a changing climate on aerosol distribution (via, for exam-4 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | ple, changes in precipitation and winds ✿✿✿✿ wind ✿ patterns, and in surface temperature, affecting deposition and gas-phase reaction rates) could be important (Pye et al., 2009;Kloster et al., 2009;Megaritis et al., 2013). This investigation, however, is beyond the scope of the present study.
We use the EMAC (ECHAM/MESSy Atmospheric Chemistry) global climate-chemistry model with the aerosol module MADE (Modal Aerosol Dynamics model for Europe, adapted for global applications). The model is able to track both aerosol mass and number concentrations, and to simulate the aerosol-cloud and aerosol-radiation interactions, hence allowing the estimation of aerosol RF effects. In addition to the global effects, we also focus on specific regions, where land-transport and shipping emissions are expected to change significantly compared to 2000, following the changes in transport activity and the implementation of various mitigation strategies and pollution-control measures.
Several recent studies dealt with aerosol distributions and related climate impacts under the RCPs. Chalmers et al. (2012) investigated the near-term (up to 2050) climate effect of aerosol in two RCP scenarios (RCP2.6 and RCP4.5, corresponding to global RFs of 2.6 and 4.5 W m −2 in 2100, respectively; see Sect. 3 for more details). They found a larger near-term net warming in RCP2.6 than in RCP4.5 even though the warming due to greenhouse gases is smaller in RCP2.6: this is due to a strong reduction in sulfur emissions and the resulting reduction in aerosol-induced cooling in RCP2.6. They also stress the key role of aerosol emissions in the quantification of near-term climate impacts. The review by Fiore et al. (2012) presented regional projections of particulate matter (PM 2.5 , i.e., particles with diameter smaller than 2.5 µm) across the 21st century, supported by the Atmospheric Chemistry Climate Intercomparison Project (ACCMIP, Lamarque et al., 2013) and the Coupled Model Intercomparison Project Phase 5 (CMIP5, Taylor et al., 2012) multi-model studies. They found an overall decline in PM 2.5 concentration for all RCPs scenarios, which does not follow the RCPs RF and CO 2 pathways: the model simulations showed that RCP6.0 and RCP4.5 represent the high and low end of the PM 2.5 range, respectively, whereas they are in the medium range of the climate projections. Takemura (2012) analysed the aerosol evolution along the RCPs using the SPRINTARS model and found that the aerosol load is 5 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | decreasing in Europe and North America, while it is still increasing in Asia due to economic growth and is expected to peak by 2050. The climate impact of aerosol was found to match the time evolution of loads, leading to reduced aerosol cooling by the end of the century and causing accelerated global warming. This further motivates the necessity of exploring the aerosol impacts from different sources in the RCPs, given that the response of short-lived species appears to be completely different from the projected changes in long-lived species and anthropogenic net RF in these scenarios. Transport-related sources are of particular relevance in this context, due to their comparatively large growth rates.
The present work adds a detailed analysis of the effects of the emissions from surfacebased transport sources on aerosol and climate under the four RCPs to the abovementioned studies. Our paper is organized as follows: the EMAC model (including the aerosol module MADE) is briefly explained in Sect. 2 (while we refer to R13 for a more detailed description). The RCP scenarios and the trends in transport-related emissions are described in Sect. 3. The impacts of land transport and shipping on aerosol surface-level concentrations and their changes between 2000 and 2030 are presented in Sect. 4. We discuss the climate impacts resulting from transport emissions in Sect. 5 and present our conclusions in Sect. 6.

Model and simulations
We apply the aerosol module MADE (Lauer et al., 2005(Lauer et al., , 2007 coupled to the ECHAM/MESSy Atmospheric Chemistry (EMAC) model (Roeckner et al., 2006;Jöckel et al., 2006). A detailed description of the EMAC-MADE system is provided in R13. Here we give just a short summary of its main features.
MADE considers 10 aerosol tracers: hydrophilic and hydrophobic black carbon (BC), hydrophilic and hydrophobic particulate organic matter (POM), aerosol sulfate (SO 4 ), nitrate (NO 3 ), ammonium (NH 4 ), mineral dust, sea salt and aerosol water. The aerosol population is described by considering three internally mixed log-normal modes (Aitken, accumulation and coarse mode) and taking into account microphysical processes such as nucle-6 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | ation, condensation (of sulfuric acid and organic compounds) and coagulation. To improve the computational efficiency, simplified gas and liquid phase chemistry schemes are adopted for this study, considering basic tropospheric background reactions (NO x -HO x -CH 4 -CO-O 3 chemistry) and the sulfur cycle.
In order to estimate the aerosol effects on climate, MADE is coupled to the cloud and radiation schemes of EMAC. Aerosol activation and interactions with clouds follow the parameterization by Abdul-Razzak and Ghan (2000) and the two-moment cloud scheme by Lohmann et al. (1999) and Lohmann (2002). Aerosol optical properties are calculated based on the Mie theory and coupled to the radiation scheme of EMAC. As in R13, the radiation and gas phase chemistry schemes of the model are decoupled, in order to isolate the aerosol impact. This is attained by prescribing radiatively active gases (ozone, CO 2 , CH 4 , N 2 O, CFC-11 and CFC-12) by means of climatologies or constant values in the radiation calculations.
Note that, although the simulations of this paper consider 2030 emissions, the ✿✿✿✿✿✿✿✿✿✿✿✿✿✿ period. This might appear inconsistent but it is a necessary choice in order to distinguish ✿✿✿ the effects of the emission changes on aerosol and aerosol-induced climate forcing from the effects of climate change on atmospheric aerosol. As mentioned above, the latter effects are not covered in this work, but will be the subject of future investigations.
For the same reason, the model simulations are run over the 1996-2005 period (plus 1spin-up) as in R13, applying the same boundary condition data and using the European Centre for Medium-Range Weather Forecast (ECMWF) data to nudge model dynamics (vorticity, divergence, temperature, and surface pressure). The adopted resolution is T42L19 for all experiments of this work, corresponding to a Gaussian latitude-longitude grid of about 2.8 • × 2.8 • , with 19 vertical hybrid σ-pressure levels from the surface to the 7 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | top model layer, which is centered at 10 hPa (∼ 30 km). A detailed evaluation of EMAC-MADE was performed by Lauer et al. (2005Lauer et al. ( , 2007 and Aquila et al. (2011), and also by R13 specifically for the regions where the impacts of the emissions from the transport sectors are known to be significant.
For each of the four RCP scenarios (2.6, 4.5, 6.0, and 8.5), we perform a reference experiment and a perturbation experiment, in which the emissions from land transport or shipping are switched off, resulting in a total of 12 simulations. The impacts of aerosol emission for each sector and scenario is ✿✿✿ are ✿ then obtained as the difference between the reference experiment and the experiment with the emissions from the corresponding sector switched off. This method (also known as 100 %-perturbation method) and the resulting uncertainties related to non-linearities in the aerosol response have been discussed in detail in R13.

2030 emissions in the Representative Concentration Pathways
Beyond concentrations of long-lived greenhouse gases, the RCP scenarios (Moss et al., 2010;van Vuuren et al., 2011a) provide future projections of the emissions of short-lived species until 2100. The emission changes of air pollutants in the RCPs are calculated taking into account the changes in driving forces (fossil fuel and fertilizer consumption), combined with both air quality control policies and climate policies. The consideration of climate policies is one of the novelty aspect ✿✿✿✿✿✿✿ aspects ✿ of the RCPs with respect to the previous SRES scenarios (Nakicenovic et al., 2000). However, as indicated below, the projections of the emissions of short-lived species are questioned whether they display consistent futures.
The four RCPs are defined according to the projected RFs and CO 2 concentrations in 2100: -RCP2.6 ( van Vuuren et al., 2007van Vuuren et al., , 2011b) follows a peak-and-decline (overshoot) pathway, characterized by a RF peak of 3 W m −2 around 2040-2050 and a decline to 2.6 W m −2 in 2100. It is the only scenario allowing for negative CO 2 emissions. 8 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | -RCP4.5 (Clarke et al., 2007;Thomson et al., 2011) is a stabilization-withoutovershoot scenario, with a nearly-constant RF value of 4.5 W m −2 between 2070 and 2100.
-RCP8.5 ) is a rising pathway, with a RF of 8.5 W m −2 in 2100.
Although the RCPs span the whole range of possible climate policies, from the most stringent RCP2.6 to the no-climate-policy RCP8.5 scenario, they do not cover the full range of air pollution projections available in the literature and do not include, for instance, a businessas-usual scenario (van Vuuren et al., 2011a). All RCPs assume a correlation between increasing income (in terms of Gross Domestic Product, GDP) and more stringent air pollution control strategies and mitigation measures. As a consequence, they are all characterized by a globally declining trend of air polluting emissions, which is not fully representative of the literature on air quality projections. This has to be considered when applying the RCPs in air quality studies, which is one of the goals of this paper. Nevertheless, we have chosen to apply the RCPs for analysing aerosol impacts in the future given their increasing use in a large number of publications (e.g., Bellouin et al., 2011;Lamarque et al., 2011;Chalmers et al., 2012;Fiore et al., 2012;Takemura, 2012;Lamarque et al., 2013;Rotstayn et al., 2013;Unger et al., 2013;Smith and Bond, 2014), and in the recent IPCC AR5 (Kirtman et al., 2013).
The emission data has a resolution of 0.5 • × 0.5 • in latitude and longitude and is divided in 12 sectors (including land transport, shipping and aviation). It includes reactive gases, aerosol and aerosol precursors, ✿✿✿✿✿✿✿✿ namely: NO x , CO, SO 2 , NH 3 , non-methane hydrocarbons (NMHCs), BC and organic carbon (OC) (Lamarque et al., 2010). Details on the processing of the emission data for use in the EMAC model and on the additional datasets used for natural emission sources (dust, DMS, secondary organic aerosol and biogenic emissions) are provided in R13.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The aerosol module MADE treats both aerosol mass and number, and therefore requires number emission fluxes to be provided as well. Emissions have to be split in the different size modes (Aitken and accumulation) of the model. To this end, assumptions on the size distributions of emitted particle are necessary and depend on the considered sector and, in some cases, on the specific aerosol compounds, as was extensively discussed in R13. In the present paper we consider only one size distribution for the particle emissions of each transport sector, following the reference cases of R13.
For land transport, we consider the number size distribution parameters recommended by Dentener et al. (2006), which is characterized by a single emission mode (corresponding to the Aitken mode of MADE), with a median diameter of 30 nm and a geometric standard deviation of 1.8. The particle emissions from ships are parametrized by means of the size distribution measured by Petzold et al. (2008) in actual ship plumes: this results in a bimodal distribution with median diameters at 70 and 260 nm, and standard deviations of 1.45 and 1.25 for the Aitken and the accumulation mode, respectively. The ratio of the total number concentrations of the Aitken mode to that of the accumulation model is about 1600 ✿✿✿✿ 134, the corresponding mass ratio is about 4.
For the analysis of the impact of land transport sector emissions on global aerosol, we consider different continental regions: North America, Europe, and Southeast Asia (top panel of Fig. 1). The goal is to quantify the regional differences in the temporal evolution of the emissions and their effects on aerosol concentrations. Differences are especially expected between the developed countries (Europe and USA) and the developing countries and fast-growing economies (India, China and other Asian countries). A similar analysis is performed for the shipping sector, focusing on specific oceanic areas, where the impact of ship emissions was found to be large in R13, namely the North Atlantic, the North Pacific and the Indian Oceans (bottom panel of Fig. 1). The results for the individual regions are then compared with the global impacts for both sectors.
The relative changes in emissions between 2000 and 2030 in the four RCP scenarios are shown in Fig. 2 for selected species and for the regions defined above. In general, all scenarios project a decrease in land transport emissions on the global scale for all species, with the exception of RCP4.5 (increase in SO 2 and small increase in BC and POM) and RCP2.6 (increase in BC and POM). When considering the different regions, it is interesting to see that land transport emissions in North America and Europe decrease in all scenarios, in view of the mitigation strategies applied in these regions. For most of the analysed pollutants, RCP8.5 is the scenario with the strongest reduction in emissions, in spite of its high long-term climate impacts. On the other hand, land transport emissions are strongly increasing in Southeast Asia, with the exception of NH 3 (all scenarios), NO x and SO 2 in RCP2.6, and BC and POM in RCP6.0 and RCP8.5.
With regard to shipping emissions, all scenarios agree in the SO 2 reduction, following the implementation of new limitations to fuel sulfur content in shipping starting from the year 2010 (Buhaug et al., 2009). The highest (smallest) reduction is projected in RCP2.6 (RCP4.5), around 80 % (20 %). The other species are expected to increase in RCP4.5 and RCP8.5, and to decrease in the other two scenarios (with the exception of NO x which is slightly increasing in RCP2.6). The pattern of changes is identical in the three oceanic regions considered for this sector. This is a consequence of the fact that the RCPs do not consider changes in the geographical distribution of shipping routes but simply apply a scaling factor to estimate future emissions. Despite the potential inconsistencies of these emission changes, their simulated effects on atmospheric aerosol and radiation are discussed in the following sections.

Impacts on aerosol surface-level concentrations and burdens
As mentioned in Sect. 2, the impacts of the emissions from a given sector on the aerosol concentrations are calculated here using a 100 % perturbation approach. For each of the scenarios, we perform a reference simulation (REF RCP ), including all emission sources, and a sensitivity simulation where the emission from either land transport (NOLAND RCP ) or shipping (NOSHIP RCP ) are switched off. The impact of, e.g., land transport ✿✿ for ✿✿ a ✿✿✿✿✿ given ✿✿✿✿✿ RCP Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | is then calculated as: (1) In the following, we will discuss the changes D in transport-induced surface-level concentrations between 2000 to 2030 for the different scenarios. For land transport, this is given by the difference: To provide a more complete view on the changes in transport-induced aerosols, we will relate this to the changes in total concentration: and to the changes in the concentrations induced by other (non-land-transport) sources: Note that, of course, RCP . An identical approach is applied to quantify the effects of shipping, using the NOSHIP instead of the NOLAND experiment in the equations above.

Land transport
The impacts of land transport emissions on large-scale mean surface-level concentrations are presented in Figs The general tendency shown by these figures is a decrease in aerosol surface-level concentrations over the developed countries (USA, Europe and Japan) and an increase in the countries with fast-growing economies (Southeast Asia). This applies to the total concentrations (as also found by Takemura (2012) using the SPRINTARS model) as well as to the land-transport-induced concentrations. Such decreasing concentrations in the developed countries is a common feature of all RCPs, but there are some exceptions, discussed in detail below. Most importantly, the changes in the aerosol concentrations are different from what could be expected from the RCP rankings: scenarios with the most stringent climate policies and with the strongest long-term reductions in global climate effects (like RCP2.6) show sometimes an opposite behaviour in terms of short-lived pollutants from specific sectors and regions.
The largest reductions of the land-transport impact on BC surface-level concentration ( Fig. 3) are simulated for Europe, in particular for RCP8.5, with local reductions up to 1 µg m −3 . Significant reductions (0.1-0.5 µg m −3 ) are also found for the USA, Japan and some large metropolitan areas (especially in South America). This result is not surprising, in view of the findings of R13 for the year 2000, in which land transport was responsible for the bulk of BC surface-level pollution in these regions (up to about 70 %). Air pollution control policies in such regions must necessarily address the land transport sector in order to reduce pollution. Compared with the changes in other sources (right column of Fig. 3), it is clear that land-transport drives most The changes in aerosol mass concentrations directly correspond to the changes in particle number concentrations (Fig. 5). This is particularly the case for primary particles, which directly increase particle number concentration. It is interesting to note, however, that ✿✿ in ✿✿✿✿✿ many ✿✿✿✿✿✿ cases ✿ land transport has only a marginal ✿✿✿✿✿✿✿✿✿✿✿✿ comparitively ✿✿✿✿✿✿ small impact on the overall changes in particle number concentrations. Such changes seem to be mostly attributable to other sources, in particular the sources emitting significant amounts of POM (Fig. S1 in the Supplement). We shall recall, however, that number emissions in the model are calculated assuming externally mixed particles (see R13, for details) and that such assumption may not always be valid for combustion processes. Changes in total concentrations are largest in RCP8.5, in particular for Europe, with decreases around 5000-10 000 cm −3 . Overall, the land-transport-induced changes in particle number follow the pattern of changes in mass, 14 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | although increases in Southeast Asia are generally quite small, especially for RCP6.0 and RCP8.5.
As discussed in R13, the estimates on particle number concentrations critically depend on the assumed size distributions of the emitted particles. The related uncertainty is particularly high for the land-transport sector, where it can be as high as one order of magnitude in the simulated local concentrations. The sign of the estimated changes, however, should be less affected by this uncertainty, as long as one can further assume that the size distribution of particle ✿✿✿✿✿✿✿✿ particles emitted by land transport does not change significantly between 2000 and 2030. This hypothesis, however, would need to be tested, considering the change in the vehicles properties in the future, which may even depend on the regions under consideration. Such an analysis, however, is beyond the scope of the present study.
To summarize the above results and further characterize the effect of the land-transport sector on air quality, we close this section with a short discussion of ✿✿✿✿✿✿✿✿✿✿ discussing the relative contributions of land transport to the aerosol near-surface burdens, depicted in Fig. 6 for the regions defined in Fig. 1 and for the globe. These plots highlight the large regional differences in the contribution of land-transport to near-surface pollution. In the year 2000, this sector accounts globally for about 20 % of the near-surface burden of BC (R13), but it has a much higher impact in Europe and North America than in Southeast Asia, where the near-surface BC pollution is mostly driven by other sources (e.g., industries and domestic heating). This ✿✿✿✿✿✿ These large regional differences are still present in 2030 in all RCPs. Given these relative contributions, Europe and North America have a particularly effective BC reduction potential from land transport, due to emission regulations and to the implementation of technological measures. However, the reduction of the land-transport impact on BC is larger in North America than in Europe, since in the latter region emission regulations are already established. In Southeast Asia, on the other hand, the land-transport share grows for most of the scenarios. The land-transport contribution to NO 3 near-surface burden, on the other hand, is nearly constant in ✿✿ for ✿ all RCPs, with the exception of RCP2.6. In term ✿✿✿✿✿ terms ✿ of particle number, the land-transport-induced burden is characterized by small variations between 2000 and 2030, for all RCPs, although there are some notable Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | exceptions. The number burden share of land-transport grows significantly in Europe under the RCP8.5 scenario, due to the large decrease of the contribution of other sources (see Fig. 5). In Southeast Asia, the contribution of land-transport to number burden is dominated by primary particles, since in this region aerosol pollution is mostly driven by primary anthropogenic aerosol.

Shipping
Our simulations reveal that the emissions from shipping have a relevant impact on the concentrations of SO 4 , NO 3 and fine particle number (Figs. 7-9, respectively, first rows), in particular in the North Pacific, North Atlantic and Indian Ocean. This also justifies the choice of the regions shown in Fig. 1 for the analysis (see also Sect. 3). Over the continents, the sulfate pollution is dominated by other sources. The impact of shipping is however non negligible The pattern of changes shows a clear decrease in shipping-induced sulfate, mostly confined to the Northern Hemisphere (Fig. 7). This is a consequence of the new regulations in ✿✿ for ✿ shipping fuel which limit the fuel sulfur content, starting from 2010 (Buhaug et al., 2009). Such regulations are accounted for in the RCPs. It is interesting to note that total sulfate is projected to decrease quite strongly in the continental regions and in the coastal areas (with the exception of southern Asia, ✿✿✿✿✿✿✿✿ southern Africa and South America) in all scenarios, but the contribution of shipping to this trend is only marginal, around 0.5 µg m −3 , whereas the concentration induced by other sources decrease by several µg m −3 . In RCP2.6, for example, the assumed climate policies to reduce greenhouse gas emissions lead also to a simultaneous reduction of short-lived pollutants. For SO 2 , in particular, the reduction in emission derives from the reduced coal use in the energy production sector (van Vuuren et al., 2011b). Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Righi et al. (2011) analysed future scenarios of lowered sulfur emissions from shipping for a set of alternative fuels (including biofuels). They simulated up to about 40-60 % reduction in shipping-induced sulfate concentration along the major shipping routes of the Indian Ocean, the northern Pacific Ocean, and the central Atlantic Ocean. A similar relative reduction is found also for the scenarios discussed here, in particular for RCP6.0 and RCP8.5, although the pattern of changes is different due to the different emission data used.
The shipping impacts on sulfate pollution are strongly correlated with aerosol nitrate effects, given that the formation of both compounds in fine particles is controlled by the availability of ammonia. The formation of ammonium-sulfate is favoured over ammonium-nitrate, therefore ammonium-nitrate can be formed only if there is an excess of ammonia available after the formation of ammonium-sulfate (Seinfeld and Pandis, 1998). Given the strong reduction in sulfate discussed above, the competition between sulfate and nitrate for available ammonia is of particular relevance for this sector. A lower concentration of SO 2 in the marine boundary layer makes more ammonia available for the formation of ammonium nitrate ✿✿✿✿✿✿✿✿✿✿✿✿✿✿✿✿✿ ammonium-nitrate and therefore results in an increase of the aerosol nitrate concentration. The results of our simulations (Fig. 8) support this line of reasoning: the increasing trend in nitrate concentration over the oceans is almost completely driven by shipping in all scenarios, with the largest effect simulated in the Northern Hemisphere. The only exception is RCP6.0, for which a slight reduction in nitrate is simulated over the Pacific. This reduction stems from a significant reduction in NO x emissions from shipping, partly compensated by a reduction in sulfate concentration.
The combined effect of SO 4 and NO 3 shows that the regulations aiming at reducing SO 2 emissions should also consider reductions in aerosol nitrate precursors (NO x and NH 3 ) in order to achieve a reduction in aerosol concentration (Kharol et al., 2013) and in aerosolinduced climate impacts (Bellouin et al., 2011). It should be noted that sulfate and nitrate could also be taken up by sea salt particles. In this case a similar competition as described above can occur between the formation of sodium sulfate and sodium nitrate. Due to their large surface area concentrationfine particles probably ✿ , ✿✿✿✿ fine ✿✿✿✿✿✿✿✿✿ particles ✿✿✿✿ can take up sulfate Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | more efficiently than the comparatively large sea salt particles. Hence, we neglected interactions of reactive gases with coarse mode aerosol in the model version applied here.
Sulfate is also central in determining the shipping effect on particle number concentration (Fig. 9). A general decreasing trend of the shipping effects is found in all scenarios, although ✿ it is less marked in RCP4.5 and RCP8.5. These two scenarios are characterized by an increase in primary particle emissions (Fig. 2, top-right panel), which contributes to number concentrations and slightly counteract the sulfate-driven reduction. In Sect. 5, we shall discuss how these changes affect the estimates of the aerosol-induced RF effects from this sector.
As for the land-transport sector, we summarize this discussion analyzing the contribution of shipping to the near-surface ✿✿✿✿✿✿✿ aerosol ✿ burden in Fig. 10. An important difference with respect to the land-transport sector is that the RCP scenarios do not implement any change in the geographical distribution of shipping emissions with respect to 2000. This is the reason why the different regions in Fig. 10 are characterized by a similar pattern of changes between 2000 and 2030. An exception is the Indian Ocean region (Fig. 10, bottom-right panel), which is strongly affected by the pollution from the Asian continent which in turn can impact the oceanic background and lead to some differences in the resulting contribution of shipping to the near-surface burden. Sulfate and nitrate are the most important contributors to near-surface pollution over the oceans, with a relative value of 14 % and 24 % in 2000 (R13), respectively. In the future scenarios, there is a general anti-correlation between these two species, due to the aforementioned competition effect for ammonia, whereby large decreases in SO 4 are counteracted by increases in NO 3 . The shipping impact on particle number concentration is generally smaller than for land transport, due to the relatively large sizes of the particles emitted by ships (Petzold et al., 2008). Globally, the relative change in this sector are mostly driven by sulfate. Given that shipping is the main source of anthropogenic pollution in the Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | marine boundary layer, even a relatively small perturbation to aerosol particle number concentration can produce a large effect on clouds and on the radiative balance, as we shall show in the next section.

Transport impacts on climate
Aerosol-induced impacts on the Earth's radiation budget due to transport emissions are quantified here as the changes in the radiative fluxes at the top of the atmosphere (cf. Lohmann and Feichter, 2005;Schulz et al., 2006). We consider the combined effect of longwave and shortwave radiation and analyse the all-sky radiative flux (considering both cloudy and clear-sky conditions) as well as the clear-sky flux only. The latter is derived from radiative flux calculations neglecting the effects of clouds and can be used to infer the importance of clouds and cloud modifications for the radiation budged (see R13, for more details). Transport-induced changes in the radiation fluxes can be interpreted as the RF of the aerosol changes caused by transport emissions. The global-mean RF obtained for the year 2000 and for the four RCPs in 2030 are presented in Fig. 11. The values for the year 2000 plotted in this figure are taken from the companion paper (R13) and correspond to the reference case, although in that paper a range of possible values was found. This range is marked by the open boxes in Fig. 11 and can be interpreted as the uncertainty in the transport-induced RF due to the assumptions on the size distribution of emitted particles. We refrain from repeating the same uncertainty analysis in the present paper, given the large number of expensive additional experiments that this would require. To give a hint of the uncertainty associated to the RF values in 2030, we assumed that the uncertainty range calculated for 2000 can also ✿✿ (in ✿✿✿✿✿✿✿✿ relative ✿✿✿✿✿✿ terms) ✿ be applied to future scenarios by simply rescaling the corresponding 2030 value according to the upper and lower values for 2000. The rescaling is given only for the all-sky values, since clear-sky values are mostly small or not significant, and hence such a scaling can be critical. However, we can still conclude that the all-sky aerosol forcing is much larger than the corresponding clear-sky value Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | for all sectors and scenarios, suggesting a dominant influence of aerosol-induced cloud modifications on the RF of aerosol changes resulting from transport emissions.
The land-transport-induced aerosol RF (Fig. 11, left panel) shows a reduction of the cooling effect from −80 mW m −2 in 2000 to about −60 mW m −2 in 2030. As mentioned above, these values represent the upper limit corresponding to the reference case(assuming AEROCOM size distribution for emitted particles). The lower limit is about a factor of 7 smaller and corresponds to assuming an aged size distribution for the emitted particles (R13 sector. The shipping impacts (Fig. 11, right panel) are characterized by a strong decrease of the aerosol-induced RF in RCP2.6, dropping from −181 in 2000 to −53 mW m −2 in 2030 as a consequence of the fuel-sulfur control measures implemented in the shipping sector, which significantly affect the sulfate burden. This factor 3-4 reduction in aerosol forcing is in good agreement with the results of Righi et al. (2011) for future low-sulfur shipping scenarios, whereas Lauer et al. (2009) simulated a factor of 2 reduction in aerosol-induced RF following a global reduction of fuel sulfur content from 2.7 to 0.5 % by mass. Smaller reductions are projected for the other scenarios, consistently with the corresponding changes in tropospheric sulfate burdens (Fig. 10). The uncertainty range due to the assumed size distribution of emitted particles is relatively small for the shipping sector, as discussed in R13.
The results in Fig. 11 show that despite a significant growth in traffic volume, the aerosolrelated climate impacts do not increase. We shall however stress again the fact that the RCP scenarios are quite optimistic in terms of air quality policies and that they do not include a business-as-usual scenario. Hence, the consideration of alternative, more realistic air-pollution scenarios may lead to different conclusions in this regard. In the case Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | of land-transport, new regulations are limiting the emissions in the developed countries, where this sector is still one of the main sources of pollution. In developing countries and fast-growing economies (e.g., southern and eastern Asia), the emissions are expected to continue growing in the near future, as discussed in Sect. 3. In these regions, however, the land-transport contribution to pollution is still limited according to our simulations. For the shipping case ✿✿✿✿✿✿ sector, the strong reduction in aerosol sulfate concentrations in the nearfuture are expected to strongly reduce the climate impact from this sector. As we ✿✿✿✿ have shown in Sect. 4.2, however, reductions in sulfate are mostly compensated to some extent by increases in aerosol nitrate, which would explain the relatively small variations in the RF effect between 2000 and 2030. A significant exception is the RCP2.6 scenario, for which such compensation effect is less evident and a ✿✿✿✿✿ strong ✿ reduction in shipping-induced particle number concentration has been simulated ( Fig. 9).
As it is clear from Fig. 11, our estimates of the climate impacts of transport sources are still affected by large uncertainties. As extensively discussed in R13, these can result from sub-grid scale particle aging effects which cannot be resolved by the coarse resolution of global models, like the one which was adopted for this study. More research on this issue is needed, by means of dedicated measurements, plume models or highly resolving regional models.

Conclusions
We simulated the effects of the global land transport and shipping sectors emissions on atmospheric aerosol and climate in the year 2030 and analyse the changes with respect to 2000, following the emission projections in the four Representative Concentration Pathways (RCP; Moss et al., 2010).
The most striking feature of the future shipping emissions is the strong reduction in SO 2 emissions projected by all scenarios and following the adoption of more stringent fuel sulfur content limits in this sector since 2010 (Buhaug et al., 2009). This leads to a clear decrease in shipping-induced aerosol sulfate concentration, especially in the Northern Hemisphere, where most of the shipping routes are located. This reduction correlates with an increase in aerosol nitrate, since the chemical processes that lead to the formations of these two compounds are controlled by the availability of ammonium. The reduction in aerosol sulfate was found to result in a significant increase of aerosol nitrate concentration and  Megaritis et al., 2013;Pozzer et al., 2014) . ✿ Both sectors are characterized by a significant reduction in the aerosol-induced RF effects, despite the growth in traffic volume. Land-transport-induced RF was found to de-22 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | crease from −80 mW m −2 in 2000 to about −60 mW m −2 in 2030. A strong decrease was simulated for shipping, from −181 in 2000 to −53 mW m −2 in 2030 for the RCP2.6, in agreement with previous assessments , while for other scenarios the decrease is less pronounced.
The analysis of changes in aerosol distribution between 2000 and 2030 frequently revealed an anti-correlation of the air-pollution impacts with the RCP ranking: the RCP scenarios with the most restrictive climate policies were often found to be characterized by a large impact in terms of short-lived pollutants, although exceptions exist. The RCPs cover a large range of future climate policies, but they have some limitations in terms of air pollution projections. They have been claimed to be excessively optimistic in terms of air pollution control, as they do not include, for example, a business-as-usual scenarios ✿✿✿✿✿✿✿✿ scenario. Air-quality policy scenarios, such as EDGAR-CIRCE (Doering et al., 2009a, b) were considered in other studies (Pozzer et al., 2012), who however focused on the total pollution in different regions, without differentiating between the different sectors, which was the major focus of the present study. Future studies on transport effects could also be based on such scenarios.
In the light of the discussion above, we shall stress that scenarios are projections of potentials futures but no forecasts. The assumptions made when constructing the emissions ✿✿✿✿✿✿✿✿ emission ✿ scenarios have a large impact on the simulation results. If the emission scenarios do not explore the full range of potential climate and air quality policies, we cannot expect that our simulation results will cover the full range of potential aerosol futures.
The present paper considered the land transport and the shipping sector. In a follow-up work, we will present ✿✿✿✿✿✿✿ perform ✿ an analogous analysis for the aviation sector, focusing on different regions and on the upper troposphere.
The Supplement related to this article is available online at doi:10.5194/acpd-0-1-2014-supplement.