We consider future changes in tropospheric ozone based on the Representative Concentration Pathways (RCPs), which are new emission and concentration scenarios for the 5th coupled model intercomparison project. In contrast to the SRES scenarios, all the RCP scenarios assume an emission reduction of NOx by the late 21st Century that has the potential to achieve tropospheric ozone reduction. However, increasing radiative forcing (RF) due to greenhouse gases and changes in CH4 concentration also contribute to differences in the tropospheric ozone distribution among RCP scenarios. In the RCP4.5 and RCP6.0, assuming the stabilization of RF, the increase in tropospheric ozone due to enhanced residual circulation is cancelled out by the ozone reduction due to ozone precursor reductions. In contrast, in the RCP8.5, assuming increasing RF even after 2100, further enhanced residual circulation and significant increase in CH4 cause a dramatic increase in tropospheric ozone.
 A substantial increase in tropospheric ozone (O3), which has been observed since preindustrial times, was driven by the dramatic increases in anthropogenic emissions of O3 precursors [e.g., Volz and Kley, 1988; Lamarque et al., 2005]. Tropospheric O3 not only has a detrimental impact on human health, agricultural crops, and ecosystems but is also considered to have the third-largest radiative forcing (RF) among anthropogenic greenhouse gases (GHGs). Therefore, it is important to investigate whether the increasing trend in tropospheric O3 during the 20th Century will persist in the future. A number of modeling studies have projected future tropospheric O3 considering future changes in climate and/or emissions [e.g., Collins et al., 2003; Sudo et al., 2003; Brasseur et al., 2006; Zeng et al., 2010]. These studies demonstrated an increase in upper tropospheric O3 in accordance with the enhanced stratosphere-troposphere exchange (STE) in the future climate and a decrease in the lower tropospheric O3 caused by increased chemical O3 loss due to water vapor in the future warmer and wetter climate. In addition, the future change in O3 precursor (NOx, CH4, CO, and Volatile Organic Compounds (VOCs)) emission is thought to have a large impact on the future tropospheric O3 [e.g., Brasseur et al., 2006]. Atmospheric chemistry models have been intercompared under a European Union project, Atmospheric Composition Change: the European Network of excellence (ACCENT) to increase the robustness of future O3 projections and quantify the uncertainty [e.g., Stevenson et al., 2006]. Most studies have, however, used the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emissions Scenarios (SRES) A1 and A2 scenarios [Nakicenovic and Swart, 2000] in the future simulations.
 Recently, a new set of emission and concentration scenarios called “Representative Concentration Pathways (RCPs)” was released for the 5th coupled model intercomparison project (CMIP5) [Moss et al., 2010]. The RCPs have four scenarios, each of which corresponds to a specific pathway towards reaching each target RF in 2100 due to long- and short-lived greenhouse gases; the target RFs are 2.6, 4.5, 6.0, and 8.5 W/m2 for the RCP2.6, RCP4.5, RCP6.0, and RCP8.5 scenarios, respectively. The RCP2.6 assumes a peak (3.0 W/m2) in the early 21st Century and a decline of RF before 2100, the RCP4.5 and RCP6.0 assume RF stabilization after 2100, and the RCP8.5 assumes an increasing RF even after 2100. These new scenarios employ different O3 precursor emissions and a wider range of GHG concentrations pathways from the SRES scenarios. In this paper, we evaluate the changes in the tropospheric O3 under the RCP scenarios using a chemistry-climate model and analyzed the factors causing them.
2. Model and Experiment
 A chemistry-climate model for the troposphere, CHASER [Sudo et al., 2002], is used in this study. The model has T42 horizontal grid spacing (approx. 2.8° by 2.8°) and 45 vertical layers from the surface up to about 80 km with ∼1 km vertical resolution in the upper troposphere and lower stratosphere. The model calculates the concentrations of 53 chemical species through 27 photolysis and 113 chemical reactions representing a detailed tropospheric chemistry involving the O3-HOx-NOx-CO-VOCs system, sulfate aerosol formation, and several heterogeneous reactions. As the model does not have halogen chemistry, the concentration of O3 above the tropopause, which is defined by the lapse rate, is nudged to the State University of New York at Albany (SUNYA) monthly mean O3 climatology data [Wang et al., 1995]. The SUNYA data are scaled by the observed stratospheric O3 trend [Randel and Wu, 1999] in the past simulation and by the projected time evolution of Effective Equivalent Stratospheric Chlorine (EESC) following the WMO98 scenario A1 [World Meteorological Organization, 1999] in the future simulation. This procedure represents a rapid decline in stratospheric O3 from the mid-1970s to 2000s and a gradual recovery thereafter, but the stratospheric O3 dataset does not take account of the impact of future climate changes on stratospheric O3. On the other hand, NOy species above the tropopause are nudged into the monthly mean data from a 3-D stratospheric chemistry model [Takigawa et al., 1999].
 Time-slice experiments were performed for every ten years from 1850 to 2100 and for 2005, as a present-day condition, under the RCP scenarios. The model was time-integrated for three years using the same boundary conditions for individual years, including a one-year spin-up. The concentrations of well-mixed GHGs (CO2, CH4, and N2O) were as prescribed by Meinshausen et al. (The RCP greenhouse gas concentrations and their extension from 1765 to 2500, manuscript in preparation, 2011), where the interaction between climate changes and CH4 lifetime was simply parameterized. Emissions of NOx, CO, non-methane VOCs (NMVOCs), and SO2 were provided by Lamarque et al.  for the past and by the RCP database (http://www.iiasa.ac.at/web-apps/tnt/RcpDb/) for the future. The HadISST dataset [Rayner et al., 2003] was used for the historical sea surface temperatures (SSTs) and sea ice concentrations (SICs). Future SSTs and SICs were prescribed using monthly anomalies of SSTs and SICs simulated with the coupled ocean-atmosphere GCM, called MIROC [Nozawa et al., 2007], added into the present-day (i.e., 2005) SSTs and SICs. We applied the SSTs and SICs from the CMIP3 experiments whose temporal evolutions of CO2 concentration are closest to that the RCP scenarios, i.e., the year 2000 commitment, SRES B1, A1B, and A2 experiments for the RCP2.6, RCP4.5, RCP6.0, and RCP8.5 scenarios, respectively. Note that the SSTs in the RCP2.6 and RCP8.5 would be underestimated since the CO2 concentrations in the RCP2.6 and RCP8.5 are slightly larger than those in the year 2000 commitment and SRES A2 experiments, respectively. Table 1 summarizes these input data for 2005 and 2100 for all RCP scenarios. Natural emissions of O3 precursors from vegetation, soil, and ocean were kept constant for 1850–2100. Neglecting a climate feedback of natural emissions, such as isoprene, may underestimate tropospheric O3 because natural emissions are affected by changes in climate and vegetation [e.g., Hauglustaine et al., 2005]. Since lightning NOx emissions are parameterized with the GCM convection scheme, they respond to the model climate as shown in Table 1.
Table 1. Summaries of GHG Concentrations, O3 Precursor Emissions, SO2, Lightning NOx, and SSTs for the Year 2005 and the Year 2100 for All RCP Scenarios
The REF shows the global mean SST in 2005, while the other RCPs indicate the SST changes from 2005 to 2100.
3. Results and Discussion
3.1. Future Projections of Tropospheric O3 Under RCP Scenarios
Figure 1 shows temporal evolutions of global mean tropospheric column O3 (TCO) calculated by the CHASER. The TCO slowly increases until 1950 and then rapidly increases until 2000. The TCO reaches 30.4 DU in 2000, which is slightly overestimated in comparison with that observed by the Global Ozone Monitoring Experiment (GOME) (28.9 DU for the 1996–2002 mean) [Liu et al., 2006]. In the 21st Century, the TCO in the RCP2.6 declines by 3.5 DU from 2005 to 2100, which is consistent with the reductions of O3 precursor emissions (Figures S1 and S2 in the auxiliary material). In contrast, the RCP4.5 and RCP6.0 have peaks of TCO around 2040 and 2070, respectively, and then the TCOs decline by 2100, resulting in nearly the same levels of TCO in 2100. The TCO in the RCP8.5 increases rapidly by 6.3 DU by 2100. These TCO changes do not correlate with the reductions of NOx and NMVOC emissions, while they are similar to the changes in CH4 concentration.
Figure 2 shows the changes in the annual and zonal mean O3 concentration and residual mean meridional circulation (hereafter, residual circulation) between 2005 and 2100 for each RCP scenario. The significant increases in the stratospheric O3 correspond to the stratospheric O3 recovery nudged in all RCP scenarios. Changes in the tropospheric O3 distribution, on the other hand, strongly depend on the individual scenarios.
 The RCP2.6 shows the O3 reduction in the entire troposphere, especially in the northern hemisphere (NH) (Figure 2a). There is no significant change in the residual circulation, but a large increase in STE is calculated (Table 2). The increase in STE is caused by the stratospheric O3 recovery and the tropospheric ozone reduction due to emission reductions. The RCP4.5 also shows a tropospheric O3 reduction. Two maxima of O3 reduction exist in the lower troposphere in the NH and around the tropical tropopause (Figure 2b), whereas the O3 increases at the altitude above 400 hPa in the mid- and high latitudes. The RCP4.5 and RCP6.0 calculate similar changes in the tropospheric O3, but the RCP6.0 calculated a more intensified tropospheric and stratospheric residual circulation (the Brewer-Dobson circulation) than in the RCP4.5; the ascent/descent in the tropics/subtropics at 100 hPa increased by about 30% and 50% in 2100 relative to 2005 in the RCP4.5 and RCP6.0, respectively. The intensified residual circulation causes enhanced O3 input from the lower stratosphere to the troposphere, penetrating more deeply into the troposphere (Figure 2c). The RCP8.5 shows a large increase in O3 that occurs in the entire troposphere except near the tropical surface and tropopause (Figure 2d). The dry deposition increases due to the increase in surface O3.The residual circulation is more intensified in the RCP8.5 than in the RCP6.0. The STE, however, is comparable to that in the RCP6.0 because of the large increase in tropospheric O3.
Table 2. Global Tropospheric O3 Budgets, Lifetimes, and TCOs for RCP Scenarios in 2100
 To investigate the factors causing tropospheric O3 changes separately, we conducted sensitivity experiments adapting the same settings of the run for the year 2005 but applied the stratospheric O3 in 2100 (STRO3), the emissions in 2100 (EMSXX), the GHGs without CH4 in 2100 (GHGXX-CH4), and all GHGs in 2100 (GHGXX-ALL). Here, XX denotes the target RF in each RCP scenario. Note that the CH4 concentration was fixed at the 2005 level in the EMSXX. We focus on the RCP2.6, RCP4.5, and RCP8.5 because the RCP6.0 shows similar results to the RCP4.5. The sensitivity experiments were time-integrated for five years, including a two-year spin-up, which is longer than those in the time-slice experiments.
 The STRO3 shows a slight increase in the TCO (31.4DU) due to an increase in STE (Table 3). The increased O3 in the lower stratosphere corresponding to stratospheric O3 recovery is transported down to the troposphere by the Brewer-Dobson circulation [Zeng et al., 2010]. The stratospheric O3 dataset in this study does not take account of future climate changes and is independent of the RCP scenarios. In the more realistic future projection, the interactions between future climate change and stratospheric O3 recovery [e.g., Eyring et al., 2010] should be considered.
Table 3. Same as Table 2 but for the Sensitivity Experiments
Dry Deposition (Tg/yr)
Note that REF shows the three-years mean under the 2005 condition for a comparison with the results of sensitivity experiments.
 The EMS2.6 and EMS4.5 show distinct decreases in the tropospheric O3, especially in the middle latitude of the NH, where many emission sources are located (Figures 3a–3b). These experiments show a large decrease in the ozone production due to emission reduction (Table 3). On the other hand, the EMS8.5 calculates small decreases in the tropospheric O3 (Figure 3c) because the RCP8.5 assumes a smaller reduction of NOx and NMVOC emissions in 2100 than the other scenarios.
 The GHG2.6-CH4 shows few changes in the tropospheric O3 (Figure 3d). Note that the underestimation of SST change in the RCP2.6 (see section 2) could result in smaller changes in residual circulation in the GHG2.6-CH4. The GHG4.5-CH4 and GHG8.5-CH4 show a decrease in the lower-tropospheric O3 and an increase in the middle- and upper-tropospheric O3 (Figures 3e and 3f), consistently with the findings of Sudo et al. . The moistening of the air due to global warming promotes the chemical O3 destruction, especially in the lower troposphere in the tropics [Sudo et al., 2003; Zeng et al., 2010]. The ozone lifetimes were reduced from 26.7 days to 24.3 days in the GHG4.5-CH4 and 22.7 days in the GHG8.5-CH4 since the moistening of the air promotes an O3 loss reaction [Stevenson et al., 2006]. In contrast, the intensified residual circulation increases the STE, contributing to an increase in the tropospheric O3. As a result, the TCOs in both experiments (31.6 DU and 33.0 DU) become larger than those in 2005 (31.0 DU).
 A change in the CH4 concentration can affect the tropospheric O3 through both radiative-dynamical and chemical processes. The GHG8.5-ALL shows a substantial increase in tropospheric O3 except around the tropical tropopause and surface (Figure 3g) despite nearly identical warming and ozone lifetime to those in the GHG8.5-CH4. The TCO (36.4 DU) is about 10% larger than the GHG8.5-CH4, where the CH4 concentration increases from 1.75 ppm to 3.75 ppm. Brasseur et al.  showed similar sensitivity of tropospheric O3 to the atmospheric CH4 level, i.e., about 13% more TCO was calculated when the CH4 concentration increased from 1.7 ppm to 3.6 ppm.
 In our experiments, the total changes in TCO in the STRO3, EMSXX, and GHGXX-ALL are slightly smaller than those in all RCP scenarios (Tables 2 and 3), which indicates that some nonlinearity exists among them.
 We evaluated future changes in tropospheric O3 on the basis of the newly released RCP scenarios for CMIP5 using the CHASER model. In the RCP2.6, the O3 precursor reductions contribute significantly to the decrease in the tropospheric O3; the global annual mean TCO in 2100 is 3.5 DU lower than that in 2005. In the RCP4.5 and RCP6.0, the increased STE due to changes in the residual circulation and O3 precursor reductions affect the changes in the tropospheric O3. Since these effects and the additional effect of stratospheric O3 recovery can roughly compensate for each other, the RCP4.5 and RCP6.0 show similar TCO levels in 2100 to that in 2005. However, two maxima of O3 reduction exist near the surface in the NH and around the tropical tropopause because of the enhanced STE, the moistening of air, and emission reductions. In contrast, the RCP8.5 shows a substantial increase in tropospheric O3; the global annual mean TCO in 2100 is 6.3 DU higher than that in 2005. The enhanced residual circulation causes a large increase in O3 in the middle and upper troposphere, and the increase in CH4 concentration plays a vital role in the dramatic increase in O3 in the entire troposphere. Hence, the CH4 control strategy is essential to reduce the tropospheric O3 as well as CO2.
 The stratospheric O3 recovery, which affects the projection of tropospheric O3, will be influenced by GHGs as well as ozone-depleting substances [Eyring et al., 2010]. Compared to our stratospheric O3 dataset, the standard CMIP5 stratospheric O3 dataset (http://pcmdi-cmip.llnl.gov/cmip5/forcing.html), which includes an impact of the GHGs, shows a decrease in the lower-stratospheric O3 in the tropics and an additional increase in the lower-stratospheric O3 in the mid-latitude in 2100. This indicates that the changes in STE and tropospheric ozone would become larger than our results if we employ the CMIP5 stratospheric O3 dataset. In addition, since the interplay among climate change, stratospheric O3 recovery, and tropospheric O3 is highly complex, a multi-model study involving more comprehensive models, such as coupled ocean-atmosphere GCMs including stratospheric and tropospheric chemistry processes, would be needed to explore this further.
 We thank the three anonymous reviewers for their helpful comments. We thank Takashi Imamura at NIES for his useful suggestions. We also thank Xiong Liu at the Harvard-Smithsonian Center for Astrophysics and Sachiko Hayashida at Nara Women's University for useful discussions and input concerning the GOME dataset. This research was supported by the Japanese Ministry of Education, Culture, Sports, Science, and Technology through the Innovative Program of Climate Change Projection for the 21st Century and partially supported by the Global Environment Research Fund (S-5-4) of the Ministry of the Environment, Japan. The SX-8R at NIES was employed to perform the simulations. The GFD-DENNOU Library was used to draw the figures.