The global distribution of tropospheric ozone (O3) depends on the emission of precursors, chemistry, and transport. For small perturbations to emissions, the global radiative forcing resulting from changes in O3 can be expressed as a sum of forcings from emission changes in different regions. Tropospheric O3 is considered in present climate policies only through the inclusion of indirect effect of CH4 on radiative forcing through its impact on O3 concentrations. The short-lived O3 precursors (NOx, CO, and NMHCs) are not directly included in the Kyoto Protocol or any similar climate mitigation agreement. In this study, we quantify the global radiative forcing resulting from a marginal reduction (10%) in anthropogenic emissions of NOx alone from nine geographic regions and a combined marginal reduction in NOx, CO, and NMHCs emissions from three regions. We simulate, using the global chemistry transport model MOZART-2, the change in the distribution of global O3 resulting from these emission reductions. In addition to the short-term reduction in O3, these emission reductions also increase CH4 concentrations (by decreasing OH); this increase in CH4 in turn counteracts part of the initial reduction in O3 concentrations. We calculate the global radiative forcing resulting from the regional emission reductions, accounting for changes in both O3 and CH4. Our results show that changes in O3 production and resulting distribution depend strongly on the geographical location of the reduction in precursor emissions. We find that the global O3 distribution and radiative forcing are most sensitive to changes in precursor emissions from tropical regions and least sensitive to changes from midlatitude and high-latitude regions. Changes in CH4 and O3 concentrations resulting from NOx emission reductions alone produce offsetting changes in radiative forcing, leaving a small positive residual forcing (warming) for all regions. In contrast, for combined reductions of anthropogenic emissions of NOx, CO, and NMHCs, changes in O3 and CH4 concentrations result in a net negative radiative forcing (cooling). Thus we conclude that simultaneous reductions of CO, NMHCs, and NOx lead to a net reduction in radiative forcing due to resulting changes in tropospheric O3 and CH4 while reductions in NOx emissions alone do not.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The long-term objective of the United Nations Framework Convention on Climate Change (UNFCCC), stated in Article 2 of the accord, is to stabilize greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system. To achieve this goal, a multitude of policies and measures that cover relevant sources, sinks and reservoirs of greenhouse gases need to be considered. The Kyoto Protocol to the UNFCCC, adopted in 1997, sets binding targets for the sum of emissions of carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6) weighted by the Global Warming Potentials (GWPs) of each gas using a 100 year time horizon. Because of their long lifetimes, these greenhouse gases are well mixed in the atmosphere, and their direct effects on the Earth's radiative balance are well quantified and understood with a high level of precision [Ramaswamy et al., 2001]. Other human-influenced chemical species in the atmosphere are not well mixed but nevertheless contribute a significant radiative forcing either directly or through their effects on other radiatively active species. Currently, no climate targets have been set for emissions of these species primarily because of the complexity in estimating their global distributions and their climate forcing. Tropospheric ozone (O3), a direct greenhouse gas and an air pollutant, is a key example. O3 is not directly emitted and its production depends nonlinearly on the emissions of its precursors (CH4, nitrogen oxides (NOx), carbon monoxide (CO), and nonmethane hydrocarbons (NMHCs)), making it difficult to determine the exact amount a country is responsible for producing. O3 and its short-lived precursors (NOx, CO, and NMHCs) modify the lifetime of CH4 by controlling the oxidative capacity of the atmosphere through reactions that produce and consume hydroxyl radical (OH). NOx, CO and some NMHCs are regulated through national air quality programs and regionally under the Convention on Long-range Transboundary Air Pollution (LRTAP). The radiative effect of O3 is partly included in the GWP for CH4, however, O3 and its short-lived precursors are not directly regulated in a climate mitigation agreement.
 Measurements and modeling studies have shown that O3 concentrations have increased significantly since preindustrial times resulting in a radiative forcing similar to that due to the increase in CH4 concentration albeit with a greater uncertainty [Ramaswamy et al., 2001]. Approximately 20% of the present CH4 radiative forcing since preindustrial times is attributed to enhanced levels of O3 associated with photochemical production from CH4. The uncertainty in total O3 forcing is mostly due to uncertainties in the preindustrial O3 level and its present distribution rather than to factors related to radiative transfer as the radiative properties of O3 are understood as well as those of CO2, CH4, N2O, and the CFCs [Berntsen et al., 2000; Gauss et al., 2003].
 The radiative forcing due to a long-lived greenhouse gas does not vary significantly (on a per molecule basis) with the location of its emissions. Therefore emission inventories that identify the quantity of each long-lived gas emitted by a particular country are sufficient to assign responsibility to that country for the forcing caused by its emissions. This is, however, not the case for tropospheric O3, a secondary pollutant that is formed photochemically in the atmosphere. The lifetime of O3, ranging from days to many weeks, is shorter than the mixing time of several months in the troposphere, resulting in a nonuniform distribution in space and time. The short lifetimes of O3 precursors and the nonlinear dependence of O3 production on precursor concentrations, particularly NOx, add to the spatial and temporal variability in O3 and make it difficult to quantify each country's contribution to the tropospheric O3 burden. Furthermore, the radiative forcing due to O3 depends on its vertical distribution, with changes in O3 concentration near the tropopause resulting in the greatest radiative forcing efficiency on a per molecule basis [Wang et al., 1980; Lacis et al., 1990; Wang et al., 1993; Forster and Shine, 1997]. Regional differences in chemical and meteorological conditions have also been shown to cause strong variations in radiative forcing from O3 [Berntsen et al., 1996; Haywood et al., 1998; Fuglestvedt et al., 1999; Mickley et al., 1999; Wild et al., 2001; Berntsen et al., 2005a]. Attributing responsibility to specific countries for radiative forcing resulting from emissions of tropospheric O3 precursors is thus not as straightforward as it is for the direct emission of long-lived greenhouse gases, because the radiative forcing depends both on the location of the precursor emissions as well as on where the O3 is formed and transported. In addition, previous modeling studies have shown that the climate sensitivity to a given radiative forcing due to a change in O3 may not be the same as for radiative forcing due to a change in CO2 [Hansen et al., 1997; Joshi et al., 2003].
 Recent discussions have focused on whether non-CO2 greenhouse agents, including O3 and aerosols, should be included in a greenhouse control strategy for climate mitigation benefits [Derwent et al., 2001; Holloway et al., 2003; Swart et al., 2004; Rypdal et al., 2005]. Hansen et al.  have noted the importance of short-lived non-CO2 greenhouse species in slowing global warming and suggested focusing on air pollutants, especially aerosols and tropospheric O3, to gain dual benefits from air pollution and climate change mitigation. Including O3 in a comprehensive climate treaty would require knowledge of the contribution of each country or region's emissions to the global O3 concentration as well as the corresponding radiative forcing. Initial efforts have been undertaken to understand and quantify these contributions. For example, Fuglestvedt et al.  investigated the response of O3 concentration and its radiative forcing due to a 20% reduction in anthropogenic NOx emissions from a few selected geographical regions. The study demonstrated that upper tropospheric O3 concentrations and the resulting O3 radiative forcing are more sensitive to NOx reductions in Southeast Asia than in midlatitude and high-latitude regions including Europe, Scandinavia, and the USA. Berntsen et al. [2005a] analyzed the impact of changes in emissions of NOx and CO individually from two regions – Europe and Southeast Asia – and showed that the global O3 burden and its radiative forcing are more sensitive to emission changes from Southeast Asia than from Europe. Some studies have examined the indirect effects of regional NOxemissions on CH4 through changes in the oxidative capacity of the atmosphere and show that the radiative forcings resulting from changes in O3 and CH4 nearly offset each other; the sign of the remaining net forcing, however, depends on the region of precursor emission reduction [Fuglestvedt et al., 1999; Wild et al., 2001; Berntsen et al., 2005a].
 Since emissions of O3 precursors from Asian countries (East Asia, Southeast Asia, the Indian Subcontinent) are rising and may continue to rise for several decades, and emissions from Africa and South America may increase in the future because of economic growth [Akimoto, 2003], we must carefully evaluate each continent/region's contribution to tropospheric O3 distributions and resulting climate forcing before including O3 or any other short-lived species in a climate mitigation treaty. Improved air quality from reduced tropospheric O3 and its precursor emissions will also benefit human health and ecosystems in addition to mitigating climate change.
 In this study, we investigate the sensitivity of global O3 and CH4 burdens and forcings to marginal reductions in regional anthropogenic surface NOx emissions alone and NOx, CO, and NMHCs emissions together from major geographic regions. We first examine NOx emissions alone as NOx is the primary limiting catalyst for O3 production [Lin et al., 1988], and it is possible to reduce NOx emissions from anthropogenic sources without significantly affecting other precursors [Heinsohn and Kabel, 1999]. We also examine marginal reductions in the combined emissions of surface anthropogenic NOx, CO, and NMHCs, because this provides an estimate of the effect of controls on the suite of O3 precursors. In a series of a three-dimensional global chemical transport model (CTM) perturbation simulations we quantify the change in tropospheric O3 burden, first, due to a 10% reduction in surface anthropogenic NOx emissions from Africa, Australia, East Asia, Europe, the Former Soviet Union, the India subcontinent, North America, South America, and Southeast Asia, and second, due to a combined 10% reduction in surface anthropogenic emissions of NOx, CO, and NMHCs from Europe, North America and Southeast Asia. We select these three regions to represent the distinct meteorological and chemical conditions associated with extratropical and tropical latitudes and the magnitude of emissions ranging from highest (North America) to lowest (Southeast Asia). We estimate the radiative forcing due to the change in tropospheric O3 concentration resulting from each of the CTM perturbation simulations using a radiative transfer model (RTM). We do not consider perturbations in anthropogenic emissions of CH4, although CH4 is an O3 precursor. Instead we quantify the increase in CH4 abundance and its radiative forcing as a result of decreases in NOx, CO, and NMHCs emissions.
 Our objective is to quantify the net effect on global radiative forcing of a marginal reduction in anthropogenic emissions of O3 precursors from each region of the world. These radiative forcings would be useful to compare the climate impact of O3 precursors with that of CO2 and would provide policymakers with a basis for crediting countries for reducing their emissions of O3 precursors, therefore lessening their impact on climate while simultaneously improving local and regional air quality. Section 2 of this paper describes the global three-dimensional CTM, MOZART-2, and the global radiative transfer model. In section 3, we describe the model simulations performed for the analysis. Section 4 presents results from MOZART-2 and RTM calculations. Uncertainties and policy implications of our results are discussed in section 5. Finally, we present conclusions in section 6.
2. Model Description
 We use the global three-dimensional chemical transport model, Model for Ozone and Related Tracers version 2 (MOZART-2) [Horowitz et al., 2003], to simulate the changes in tropospheric O3 concentration resulting from a 10% reduction in continental anthropogenic emissions, either of NOx alone, or of combined NOx, CO and NMHCs. MOZART-2 simulates the distribution of 63 chemical species from the surface to the lower stratosphere (4 mb) with a horizontal resolution of 2.8° latitude × 2.8° longitude with 34 hybrid vertical levels, using a time step of 20 minutes for all chemical and transport processes. In this study, we drive MOZART-2 with meteorological variables from the middle atmosphere version of the Community Climate Model (MACCM3) archived every 3 hours. A full description and evaluation of the version of MOZART-2 used here are given by Horowitz et al. .
 Built on the framework of the transport model MATCH (Model of Atmospheric Transport and Chemistry [Rasch et al., 1997], MOZART-2 accounts for advection, convection, boundary layer transport, surface emissions, photochemistry, and wet and dry deposition. The flux-form semi-Lagrangian advection scheme of Lin and Rood  is used for tracer transport. Convective mass fluxes are rediagnosed using the Hack  scheme for shallow and midlevel convection and the Zhang and McFarlane  scheme for deep convection. Studies have demonstrated that changes in ozone concentration near the tropopause result in the greatest greenhouse forcing efficiency on a per molecule basis [Wang et al., 1980; Lacis et al., 1990; Wang et al., 1993; Forster and Shine, 1997]. The vertical distribution of ozone and its precursors, particularly short-lived NOx species, is determined to a large extent by convective transport of surface pollutants from the boundary layer to the upper troposphere on short (hourly) timescales. Therefore, to evaluate the sensitivity of global radiative forcing to regional air pollution, it is important to accurately represent convective transport. The combination of the Hack  and Zhang and McFarlane  schemes have been shown to provide a realistic transport of trace species from the surface to the upper troposphere [Rasch et al., 2003].
 In MOZART-2 the stratospheric concentration of O3 is constrained above the local thermal tropopause (defined by a lapse rate of 2 K km−1) by relaxation toward observed O3 climatologies [Randel et al., 1998; Logan, 1999]. Surface emissions of chemical species are taken from the MOZART-2 emission inventory used by Horowitz et al.  which is intended to represent the early 1990s. Surface emissions of NOx include emissions from fossil fuel burning and other industrial activities, biofuel combustion (including fuelwood burning), biomass burning, and biogenic emissions from vegetation and soils. NOx emissions from fossil fuel and biofuel combustion are based on the Emission Database for Global Atmospheric Research (EDGAR) v2.0 inventory [Olivier et al., 1996]. The global total NOx emission is 40.8 Tg N/yr with fossil fuel combustion and industrial activities contributing 23.1 Tg N/yr and biofuel consumption resulting in 1.3 Tg N/yr. MOZART-2 also includes aircraft and lightning emissions of NOx accounting for 0.67 Tg N/yr and 3.0 Tg N/yr, respectively. MOZART-2 simulates the emissions and chemical losses of CH4; however, the adjustment timescale for CH4 is sufficiently long (approximately 12 years) that CH4 does not reach a new steady state within the 2-year simulations conducted here. CH4 concentrations are thus sensitive to the initial conditions used, which are from a previous simulation of MOZART and agree reasonably with observations. The longer-term adjustment of CH4, dominated by the tropospheric “primary mode,” is treated in section 4.2.
 MOZART-2 driven by MACCM3 climatology has been extensively evaluated by comparing results with observations from ozonesondes, aircraft, and surface-monitoring stations and has been shown to simulate the concentrations of tropospheric O3 and its precursors reasonably well [Horowitz et al., 2003]. A realistic simulation of the distribution of NOx is particularly important for this study because of the nonlinear photochemistry. Comparison of model-simulated vertical profiles of NOx concentrations with observations from field aircraft campaigns show that MOZART-2 simulates NOx very well at almost all locations, over a range of concentrations spanning several orders of magnitude [Horowitz et al., 2003].
2.2. GFDL Radiative Transfer Model
 Radiative forcing from simulated changes in ozone concentration due to a 10% reduction in regional anthropogenic emissions of O3 precursors is calculated using the Geophysical Fluid Dynamics Laboratory (GFDL) global three-dimensional radiative transfer model (RTM). The RTM is a component of the new global atmosphere (AM2) and land surface model (LM2) developed at the GFDL for climate research [GFDL Global Atmospheric Model Development Team, 2004]. The RTM performs solar and terrestrial radiative transfer calculations. The solar radiative transfer algorithm follows the two-stream δ-Eddington multiple band parameterization of Freidenreich and Ramaswamy . The solar spectrum ranges from 0.2 to 4 μm and is divided into 18 bands to account for the absorption by CO2, H2O, O2, and O3, molecular scattering, and scattering and absorption by aerosols and clouds. Results from the solar radiation code have been tested against benchmark calculations using the HITRAN catalogue and the maximum error in the clear-sky heating rates is less than 10% [Freidenreich and Ramaswamy, 1999]. The terrestrial radiative algorithm is based on the modified version of the Simplified Exchange Approximation (SEA) method developed and evaluated by Schwarzkopf and Ramaswamy . The terrestrial spectrum ranges from 4.55 μm to ∞ to account for the absorption and emission by major atmospheric gases, including H2O, CO2, N2O, O3, CH4, and the halocarbons CFC-11, CFC-12, CFC-113 and HCFC-22. Absorption by aerosols and clouds in the longwave is also considered.
 In previous studies, the RTM has been used to assess the radiative forcing due to both well-mixed greenhouse gases [Schwarzkopf and Ramaswamy, 1999] and short-lived forcing agents (O3 and aerosols) [Haywood and Ramaswamy, 1998; Haywood et al., 1998]. In addition, Haywood et al.  show that the radiative forcing due to human-induced changes in tropospheric O3 calculated using the off-line RTM and the full GCM are in reasonable agreement, suggesting that monthly mean climate variables may be used without introducing significant biases in the calculated radiative forcings.
 In the present study, the RTM simulations use archived meteorological fields including insolation, temperature, specific humidity, cloud amount, and surface reflectance that are simulated by AM2/LM2 for the early 1990s. The RTM is run for each grid column of AM2/LM2, which has a horizontal resolution of 2.5° longitude × 2° latitude with 24 vertical levels from the surface to 3 mb. Random cloud overlap is assumed in the model. Concentrations of the well-mixed greenhouse gases are set at values typical of the early 1990s. The tropopause is assumed to vary linearly with latitude from a pressure of 100 mb at the equator to 300 mb at the poles and is zonally invariant (“linear tropopause”).
3. Numerical Simulations
3.1. O3 Concentration Simulations
 We perform a “base case” simulation of MOZART-2 with the standard emission inventory described earlier to obtain a reference chemical state. Similarly, we perform a series of perturbation simulations described below that provide a quantitative estimate of the O3 changes resulting from potentially feasible reductions in regional emissions of anthropogenic NOx or combined NOx, CO, and NMHCs emissions. All simulations are run for 25 months and results for the last 12 months are used for analysis.
 We perform MOZART-2 simulations in which surface anthropogenic NOx emissions from each of the nine regions shown in Figure 1 are reduced by 10%. Similarly, we perform MOZART-2 simulations with combined anthropogenic emissions of NOx plus CO and NMHCs reduced by 10% for Europe, North America and Southeast Asia. Because of computational constraints, we limit our analysis for combined NOx, CO, and NMHC reductions to these three regions that represent the distinct meteorological and chemical conditions associated with extratropical and tropical latitudes. Previous modeling studies have suggested that a 10% perturbation in NOx emissions is small enough to avoid significant nonlinear chemical influence on O3 production and large enough to produce a measurable response in O3 [Wild and Akimoto, 2001; Kunhikrishnan and Lawrence, 2004]. The difference between the global O3 distributions in a perturbation simulation and the base case indicates the effect of reductions in a region's precursor emissions. Each region's anthropogenic NOx emissions in the base case is presented in column 2 of Table 1a and anthropogenic CO and NMHC emissions are presented in columns 2 and 3 of Table 1b, respectively. The highest anthropogenic NOx emissions are from North America (NA) while the lowest are from Australia (AU). Since emissions from each of these regions are reduced by 10% from the base emissions in the perturbation simulations, the magnitudes of reduction vary across the regions.
Table 1a. Global Reductions in O3 Burden and Sensitivity of O3 Reductions to Regional Reductions in Anthropogenic NOx Emissionsa
Anthropogenic NOx, Tg N yr−1
Total NOx, Tg N yr−1
ΔO3, Tg yr−1
ΔO3/ΔENOx, Tg/Tg N yr−1
ΔCH4/ΔENOx, ppb/Tg N yr−1
Columns 2 and 3 show base surface anthropogenic and total NOx emissions (Tg N yr−1) for each region, respectively. Columns 4 and 5 show the simulated changes in global total O3 burden (Tg) associated with short-lived and primary modes (as described in section 4.2), respectively, resulting from a 10% reduction in regional anthropogenic NOx emissions. Columns 7 and 8 show the estimated changes in CH4 lifetime and the steady state CH4 concentration. Changes in O3 and CH4 burden normalized with respect to the regional NOx emission reductions are shown in Columns 9 and 10, respectively.
Africa and Middle East (AF)
East Asia (EA)
Former Soviet Union (FSU)
Indian Subcontinent (IN)
North America (NA)
South America (SA)
South East Asia (SE)
Table 1b. Same as in Table 1a but for a Combined 10% Reduction in Regional Anthropogenic NOx, CO, and NMHC Emissionsa
Anthropogenic CO, Tg yr−1
Anthropogenic NMHC, Tg C yr−1
ΔO3, Tg yr−1
ΔO3/ΔENOx, Tg/Tg N yr−1
ΔCH4/ΔENOx, ppb/Tg N yr−1
Columns 2 and 3 show base regional surface anthropogenic CO and NMHC emissions.
North America (NA)
Southeast Asia (SE)
3.2. Radiative Flux Calculations
 Since MOZART-2 does not simulate O3 concentration changes in the stratosphere, monthly mean tropospheric O3 concentrations for the base and perturbed simulations described above are merged with the observed stratospheric O3 values for 1990 (data compiled from SAGE I + II and ozonesondes by W. Randel (personal communication, 2004)) and then interpolated to the RTM horizontal and vertical grid. The O3 concentrations are thus only allowed to vary below the linear tropopause and any changes that occur above the linear tropopause are neglected in the RTM. The base and perturbed O3 distributions are used to perform RTM simulations with meteorological fields from AM2/LM2 sampled for one day per month at midmonth to represent monthly mean conditions. The monthly mean net irradiance (solar and terrestrial) at the tropopause is calculated for the base and perturbed O3 distributions. “Instantaneous” radiative forcings are calculated as the difference in the net irradiance at the tropopause between the perturbed case and the base case. “Instantaneous” here implies that we do not allow the stratospheric temperature to adjust to equilibrium after perturbing the surface-troposphere energy budget to account for the changes in the infrared emission from the stratosphere to the troposphere in the radiative forcing [Intergovernmental Panel on Climate Change (IPCC), 1995; Fuglestvedt et al., 2003]. For some climate mechanisms, such as changes in stratospheric ozone, the distinction between adjusted and instantaneous radiative forcing is crucial as they can be of opposite signs [Fuglestvedt et al., 2003]. For changes in tropospheric O3, Haywood et al.  show that stratospheric adjustment will reduce the instantaneous radiative forcing by only about 10%. Since any changes in O3 that occur above the linear tropopause are neglected in our MOZART-2 and RTM simulations (see above), stratospheric adjustment is likely to make only a small difference in our results and is unlikely to change the sign of instantaneous radiative forcing calculated here.
4.1. Ozone Perturbations Resulting From Regional Emission Reductions
4.1.1. Effect of Anthropogenic NOx Emission Reduction
 Reduced emissions of NOx, the limiting catalyst in O3 production in the free troposphere and the marine boundary layer, cause a reduction in O3 concentration in all but urban areas and high-NOx plumes. The magnitude and spatial distribution of the O3 reduction, however, varies with season, the region from which NOx emissions are reduced and the background levels of NOx, CO and NMHCs. The vertical distribution of the annual mean O3 reduction depends on the dynamical regime of the region from which the emissions are reduced (Figure 2). NOx reductions from tropical regions, including Southeast Asia, the Indian subcontinent, and East Asia result in pronounced O3 reductions (75–150 pptv) in the middle to upper troposphere that are largely confined to the latitude range of the source region. These tropical regions are characterized by deep convective activity that results in rapid transport of O3 and its precursors from the boundary layer to the upper troposphere where ozone production efficiency is higher. Reductions in anthropogenic NOx emissions from Africa and South America also result in O3 decreases in the middle troposphere; however, the reductions are smaller and spread out because these source regions do not have strong convection so O3 and its precursors remain at lower altitude. NOx reductions from midlatitude regions, including Australia, Europe, the Former Soviet Union and North America result in the largest decreases in O3 in the midtroposphere (2–8 km) with perturbations extending poleward. The largest reductions in O3 results from 10% reductions in NOx emissions from North America because the absolute magnitude of emissions from North America is the largest.
Figure 3 shows the change in the annually averaged tropospheric O3 column due to 10% NOx emission reductions for each of the nine regions. The tropospheric O3 column is calculated from the surface up to the model vertical level at which O3 concentration equals 150 ppbv in the base simulation following Prather et al. . Short lifetimes of O3 and its precursors and regional differences in transport timescales result in significant spatial gradients in the O3 perturbations. The largest column reductions occur locally near the source of the emissions; however, the plume of reduced O3 column extends zonally. For North America and East Asia, the regions of reduced O3 column extend across the Atlantic and Pacific Ocean, respectively. The horizontal long-range transport of O3 and its precursors from these regions has been shown to be efficient especially in spring and autumn when the boundary layer O3 production and vertical lifting are greater than in winter [Wild and Akimoto, 2001]. For Europe, Former Soviet Union and Australia, the reduced O3 regions are less diffuse. The perturbed O3 column shows a dipole over Europe, with a small increase in O3 column over northwestern Europe but a reduction over Eastern Europe. Small reductions in industrial NOx emissions from polluted areas of northwestern Europe result in less wintertime titration of O3 as shown by Wild and Akimoto  and cause O3 columns to increase (up to 5 × 10−2 DU). Locally, the peak O3 column decrease from emission reductions in South America, Southeast Asia, and the Indian subcontinent are similar to that from North America (0.4–0.6 DU), although anthropogenic NOx emitted from these regions is less than half that emitted from North America (Table 1a). High photochemical and convective activity throughout the year accompanied by relatively high VOC/NOx concentration ratios resulting in part from biomass burning lead to more efficient O3 production in tropical than midlatitude regions. The changes in global O3 burden range from −0.23 Tg for North America to −0.02 Tg each for Europe and Former Soviet Union (Table 1a and Figure 4a). The sensitivity of global O3 burden to a region's NOx emissions, calculated as the normalized change in global O3 burden per unit change in NOx emissions (ΔO3/ΔENOx), is highest for low NOx emitting tropical regions (Southeast Asia, South America and Australia) and lowest for high NOx emitting midlatitude and high-latitude regions with concentrated emissions (Europe, the Former Soviet Union; Figure 4b and Table 1a). The sensitivity of global O3 change to NOx reduction from Southeast Asia is almost 9 times higher than that from North America (Figure 4b). The sensitivities calculated for NOx emission reductions from the Indian subcontinent and East Asia are lower than for other tropical regions because of higher background NOx emissions from these regions.
 There are large differences in the seasonal cycle of the O3 perturbations. The monthly variations in global O3 changes are shown for both the surface (Figure 5a) and upper troposphere (approximately 12 km; Figure 5b). Surface O3 is important for air quality issues while upper troposphere O3 is important for climate forcing (see section 4.3). The seasonal cycle in O3 perturbation is driven by the seasonality in photochemistry and convective activity with a maximum during summer (Figure 5a). In the Northern Hemisphere, global surface O3 is reduced by 0.03 to 0.07 ppb during summer months because of the 10% NOx reductions from extratropical regions (North America, Europe, Former Soviet Union, and East Asia; Figure 5a). The summer peak is attributed to high photochemical activity and O3 production efficiency as shown in Figure 5c [Hirsch et al., 1996]. Within the United States, O3 production is primarily NOx limited during summer and becomes HOx limited in other seasons [Jacob et al., 1995]. During winter the O3 perturbation is small because of low photochemical activity. Reductions in NOx emissions from Europe and the Former Soviet Union result in an increase in O3 due to less wintertime titration of O3 by NOx particularly near the source of precursor emissions. There is negligible seasonality in surface O3 for NOx reductions from the Indian subcontinent, Southeast Asia, and South America because these equatorial regions experience relatively constant insolation throughout the year. The surface O3 reductions are smaller than those for midlatitude regions because of lower NOx emissions, resulting in higher sensitivity to NOx reductions as shown in Figure 5c. It is difficult to analyze the seasonality of O3 reduction from Africa because anthropogenic NOx emissions are quite small and are located in both the northern and the Southern Hemispheres. Furthermore, Africa is strongly influenced by emissions from biomass burning with peak intensities during January–March north of the equator and August–November in the Southern Hemisphere. Lower NOx emissions combined with emissions of precursors from biomass burning for Africa increase the VOC/NOx concentration ratio resulting in enhanced O3 sensitivity at the surface compared with sensitivities to NOx reductions from North America, Europe and the Former Soviet Union (Figure 5c). The highest sensitivity to NOx reduction is simulated for Australia where anthropogenic NOx emissions are the smallest. A cleaner environment and high VOC/NOx ratio due to biomass burning from May to October make this region strongly NOx limited.
 The seasonality in global O3 reductions is enhanced in the upper troposphere, particularly for North America, East Asia, and India (Figure 5b). For Europe and the Former Soviet Union, O3 reductions at 12 km appear to be close to zero throughout the year (in part because 12 km lies in the stratosphere at these latitudes). Convection over Europe and the Former Soviet Union (particularly in the western part where most anthropogenic emissions are located) is weak and O3 production and destruction occurs mostly in the lower to middle troposphere (Figure 2). For North America and East Asia, vertical transport is important during summer resulting in large O3 reductions over these regions in the upper troposphere. O3 reductions are enhanced in the upper troposphere as compared to the surface for NOx emission reductions from Southeast Asia, South America and the Indian subcontinent and show seasonality associated with convective activity. The Indian subcontinent, particularly, shows significant O3 reduction during the summer monsoons when deep convection lifts the pollutants to high altitudes where low-NOx conditions result in high O3 production efficiencies [Berntsen et al., 1996; Wild and Akimoto, 2001]. Small O3 reductions are simulated for Australia and Africa at upper levels.
 We find global O3 reductions in the upper troposphere to be most sensitive to NOx emission reductions from tropical regions (Southeast Asia, South America and the Indian subcontinent), with highest sensitivity calculated for Southeast Asia (Figure 5d). We find low sensitivities for NOx emission reductions from midlatitude and high-latitude regions with lowest sensitivity for Europe and the Former Soviet Union. The sensitivities of upper tropospheric O3 to NOx emissions calculated here are consistent with the study of Fuglestvedt et al.  who reduced anthropogenic NOx emissions by 20% and found the sensitivity to be highest for Southeast Asia followed by Australia and lowest for Scandinavia.
4.1.2. Effect of Anthropogenic NOx, CO, and NMHC Emission Reduction
 Annually averaged O3 columns are reduced as a result of combined 10% reductions in anthropogenic emissions of NOx, CO, and NMHCs for Europe, North America and Southeast Asia (Figure 6 and Table 1b). For the three regional emission perturbations, reducing CO and NMHCs in addition to NOx slightly reduces the spatial inhomogeneity in the perturbed O3 column because of the relatively longer lifetime of CO (approximately 2 months) compared with NOx (compare Figures 6 and 3). The general spatial pattern of regional horizontal gradients, however, is still the same with higher localized column reductions. Compared with NOx emission reduction only, the maximum reduction in column O3 from combined reductions in NOx, CO, and NMHC emissions is enhanced by 0.06, 0.05, 0.03 DU for North America, Europe and Southeast Asia, respectively. For Europe, combined reductions in emissions of NOx, CO, and NMHCs result in an overall reduction in O3 column in contrast to the small increase simulated over northwestern Europe for NOx emission reductions alone. The total reduction in O3 burden for the combined emission reductions is largest for North America (Table 1b). We expect that combined reductions in emissions of NOx, CO, and NMHCs from each of the remaining six regions of the world would reduce the global O3 burden; however, the distribution of O3 reductions would depend on the location and chemical regime of the source region.
4.2. Methane Increases Resulting From Regional Emission Reductions
4.2.1. Effect of Anthropogenic NOx Emission Reductions
 A reduction in NOx emissions decreases the oxidizing capacity of the atmosphere by decreasing hydroxyl radical (OH) concentrations and therefore increases the lifetime of CH4 in the atmosphere [Prather, 1994]. The lower OH results from a decreased source of odd-hydrogen (HOx = OH + HO2 + RO2) radicals (from photolysis of O3 followed by reaction of O(1D) with water vapor), and from decreased recycling of peroxy radicals back to OH. In addition, there is a positive feedback between CH4 and OH by which the increase in CH4 causes OH to decrease further, increasing CH4. Reductions in O3 and OH resulting from NOx reductions are readily captured in 2 year MOZART simulations because of the short adjustment times for these species (“short-lived mode”). Perturbations in CH4 induced by changes in OH, however, approach steady state with an e-folding time of approximately 12 years [Prather, 1994, 1996; Derwent et al., 2001; Prather et al., 2001; Wild et al., 2001]. As CH4 increases toward its new steady state (not captured in our 2-year CTM simulations), the peroxy radical production rises, enhancing O3 production in response to this “primary mode” of the tropospheric photochemistry system [Prather, 1996; Wild and Prather, 2000; Derwent et al., 2001; Wild et al., 2001].
 Calculating the perturbed steady state CH4 concentration would require significantly longer and more expensive simulations of MOZART than performed here. Fuglestvedt et al. , however, suggested a method to calculate the steady state CH4 perturbations on the basis of the initial changes in CH4 lifetime calculated from the shorter simulations. Using this method, we estimate the steady state CH4 concentration change Δ[CH4] as:
where τ0 is the lifetime of CH4 versus reaction with tropospheric OH in the base simulation which is 9.0 yr in our model, Δτ is the change in CH4 lifetime for the perturbation simulations, [CH4]0 is the CH4 concentration in the base simulation and FCH4 is a feedback factor that quantifies the positive feedback between CH4 and OH described above and is expressed as the ratio of adjustment time to lifetime of CH4 [Schimel et al., 1996; Ramaswamy et al., 2001]. FCH4 is model-dependent and its calculation requires expensive multidecadal simulations of MOZART-2 which are not currently feasible. We therefore use the value of 1.4 recommended by Prather et al. . This long-term increase in CH4 leads to an increase in O3, which partially offsets the NOx-induced decrease in O3 described above. To quantify this increase in O3, we use results from the OXCOMP experiment [Prather et al., 2001; Gauss et al., 2003], following Berntsen et al. [2005a]. In OXCOMP, the change in global mean tropospheric O3 in response to a 10% increase in CH4 was calculated by six global 3-D CTMs to be 0.64 DU (overestimated by about 25 to 33% as it includes some contribution from O3 changes in the lower stratosphere [Prather et al., 2001]). We use this average O3 response to estimate (ΔO3)primary, on the basis of the steady state CH4 change Δ[CH4] as:
 Columns 4 and 5 of Table 1a show the perturbation in O3 burden from the short-lived mode (immediate reduction in O3 from NOx emissions reduction) and long-lived primary mode (increase in O3 caused by an increase in CH4 at steady state), and the total change in O3 burden as a sum of the short-lived and primary modes is shown in column 6. As shown in columns 7 and 8 of Table 1a, there are regional differences in the response of CH4 lifetime and its steady state concentration to a 10% reduction in NOx emissions. We have estimated the global, annual mean change in steady state CH4 resulting from perturbed emissions; however, this perturbation in CH4 has a spatially and temporally varying distribution as noted by previous studies [Wild and Prather, 2000; Derwent et al., 2001]. The largest increase in the steady state concentration of CH4 is for NOx reductions from North America and the smallest for the Former Soviet Union. The normalized CH4 changes per unit NOx emissions (column 10 of Table 1a) indicate that as for O3, CH4 change is most sensitive to changes in NOx emissions from low-NOx regions (Southeast Asia, South America and Australia) and least sensitive to high-NOx regions (Europe and the Former Soviet Union). The least sensitive regions are also characterized by low photochemical activity.
4.2.2. Effect of Anthropogenic NOx, CO, and NMHC Emission Reductions
 The long-lived primary mode changes in O3 and CH4 are diminished when regional CO and NMHCs emissions are reduced in addition to NOx emissions (columns 5 and 8 of Table 1b). Reaction with OH is the primary loss mechanism for CO and NMHCs in the atmosphere; hence reduced CO and NMHC emissions tend to increase OH, opposing the NOx-induced decrease in OH. For emission reductions from Europe, CO and NMHC reductions offset the effect of NOx reductions, resulting in a change in sign (decreases) of the steady state CH4 concentration change and the primary mode O3 change (decrease) compared with NOx reduction alone. For emission reductions from North America and Southeast Asia, a net increase in CH4 remains resulting in an increase in the primary mode O3, although, only by about half as much as the increase from NOx emission reductions alone.
4.3. Radiative Forcing Due to Perturbed Ozone and Methane
4.3.1. Effect of Regional Anthropogenic NOx Emission Reductions
 The annual instantaneous total-sky radiative forcing due to short-lived O3 changes for the simulations with reduced NOx emissions is shown in Figure 7. The spatial pattern of the forcings mostly reflects the distribution of annual O3 column change simulated for each region (Figure 3), but the horizontal gradients in the forcings are stronger than in the O3 column changes. The region of maximum forcing reduction for each simulation is located near the source of NOx emissions and the region of maximum O3 reduction. There is a systematic shift in the maximum radiative forcing reductions toward lower latitudes because of the larger contrast between surface and upper tropospheric temperatures, in the tropics versus at the poles [Haywood et al., 1998]. The reduced forcing extends to other regions following the plume of O3 reductions. For example, the forcing from North America extends across the Atlantic into Europe and northern Africa. Emission reductions from most regions induce forcing reductions in both hemispheres except for reductions from the Former Soviet Union, Europe, and Australia for which reduced forcing occurs only in their respective hemispheres. Locally, maximum negative radiative forcing from Southeast Asia, South America and the Indian subcontinent is similar to that from North America. Significant O3 reductions occur near the tropopause for emission reductions from these tropical regions (Figure 2) where O3 is most radiatively effective leading to large forcings from these regions. The dipole pattern simulated for the O3 column for emission reductions from Europe is not replicated for the radiative forcing reduction, possibly because the O3 column change is determined by increases in the boundary layer while radiative forcing is influenced by changes occurring near the tropopause.
Figure 8 shows the monthly variation in the global average O3 radiative forcing from regional NOx emission reductions. The seasonal cycle in the simulated radiative forcings is due to the monthly variation in the vertical distribution of O3 perturbations and can be explained by the seasonality in the global O3 reduction in the upper troposphere (Figure 5b). Seasonally uniform O3 changes in the upper troposphere resulting from NOx emission reductions from tropical regions (Southeast Asia, South America, and Africa) result in nearly constant yearly forcings while emission reductions from high-latitude and midlatitude regions (North America, Europe, the Former Soviet Union) result in peak negative radiative forcings during summer months consistent with maximum upper tropospheric O3 perturbations simulated during summer. Emission reductions from the Indian subcontinent and East Asia also result in peak radiative forcing reductions during summer associated with O3 reductions in the upper troposphere resulting from strong convective activity.
 The global annual radiative forcings due to O3 and CH4 changes are summarized in Figure 9a and Table 2, and the normalized forcing per unit NOx (ΔFO3/ΔENOx) is shown in Figure 9b. The radiative forcing for O3 in Table 2 includes contribution from the short-lived (described above) and long-lived primary modes. We calculate the primary mode O3 forcing for each perturbation by first calculating the global average O3 forcing per Dobson unit, on the basis of the short-lived mode changes averaged over all emission reduction regions (0.045 Wm−2/DU). This forcing efficiency is then scaled by the primary mode O3 column change estimated for each emission reduction region. Our calculated average short-lived mode O3 forcing per Dobson unit change lies within the range of 0.033 to 0.056 Wm−2/DU given by Ramaswamy et al. . The strongest global O3 radiative forcing from NOx emission reductions is for Southeast Asia (−1.24 mWm−2), with the weakest for Europe and the Former Soviet Union (−0.03 mWm−2). The O3 radiative forcing per unit NOx is 15 times higher for Southeast Asia than for North America (Figure 9b). Fuglestvedt et al.  calculated the sensitivity of O3 forcing to NOx emission reductions to be highest for Southeast Asia and lowest for Scandinavia and the USA, without accounting for the primary mode effect on O3 concentrations. Their calculated O3 radiative forcing per unit NOx reduction was about 7 times higher for Southeast Asia than for the USA, while our calculated O3 forcing sensitivity is 11 times higher for Southeast Asia than for North America O3 (short-lived mode only). Differences in the definition of regions and the simulated vertical O3 distribution can possibly explain the differences in our results compared with the results of Fuglestvedt et al. .
Table 2. Global Instantaneous Cloudy-Sky Radiative Forcings Due to Changes in Global O3 (Short-Lived Mode Plus Primary Mode) and CH4 Concentrations Resulting From a 10% Regional Reduction of Anthropogenic NOx Emissions for the Nine Geographical Regions Considered in the Study
ΔF O3, mWm−2
ΔF CH4, mWm−2
Net ΔF, mWm−2
Results from combined reductions in NOx, CO, and NMHCs.
 NOx reductions enhance the lifetime and burden of atmospheric CH4 as described in section 4.2.1. We calculate the radiative forcing due to CH4 increases resulting from 10% NOx emission reductions for the nine simulations on the basis of the CH4 changes in Tables 1a and 1b using the simple formulation described by IPCC . We use this formulation because we cannot conduct MOZART-2 simulations that fully account for the CH4 concentration increases in response to emission reductions. The results are shown in Table 2 and Figure 9a. The CH4 forcing is largest for NOx emission reductions from North America and smallest for the Former Soviet Union. The net radiative forcing from O3 and CH4 changes is positive for NOx emission reductions from all regions (column 4 of Table 2). This indicates that reducing NOx emissions alone, from any region in the world, results in a small net warming. Our results are consistent with previous studies that have either shown or suggested a similar offsetting effect of CH4 to NOx-induced changes in O3 [Fuglestvedt et al., 1999; Kheshgi et al., 1999; Mayer et al., 2000; Derwent et al., 2001; Wild et al., 2001; Wigley et al., 2002; Berntsen et al., 2005a].
4.3.2. Effect of Regional Anthropogenic NOx, CO, and NMHC Emission Reductions
 Total O3 forcing (short-lived plus primary mode) becomes more negative when CO and NMHC emissions are decreased in addition to NOx, while the increase (or in one case decrease) in forcing due to CH4 changes is less (see the three bars in the right corner of Figure 9a and Table 2). The largest negative global O3 radiative forcing is simulated for North America and the smallest for Europe. The sensitivity of O3 to NOx emissions is further enhanced for Southeast Asia when CO and NMHC emissions are also reduced (Figure 9b). The combined emissions reductions from the regions considered here lead to a net negative global radiative forcing, indicating a net cooling. These results are conceptually consistent with the modeling study of Wild et al.  who demonstrated that combined increases in NOx and CO yields a net positive radiative forcing due to O3 and CH4 changes, while increases in NOx alone result in a net negative forcing with the magnitude depending on the region.
 Overall, MOZART-2 simulates the global distributions of O3 and its precursors reasonably well. However, MOZART-2 has been shown to overestimate O3 in the upper troposphere at middle to high northern latitudes, possibly because of excess stratosphere to troposphere exchange [Horowitz et al., 2003]. This and other biases in the base simulation may introduce uncertainties in our results. We have not considered changes in biomass burning emissions, even though these are mostly controlled by humans particularly in the tropics. The effects of biomass burning emissions will be addressed in a future study (V. Naik et al., manuscript in preparation, 2005). We have also not applied emission perturbations to aircraft sources that may have larger impacts on upper tropospheric O3 from direct emissions in the region.
 A direct comparison of the global O3 and CH4 forcings does not give a realistic picture of potential climate effects of O3 precursors because both O3 and CH4 respond to changes in emissions on very different temporal and spatial scales [Wild et al., 2001]. O3 and its forcing respond on a very short timescale (several weeks) via the short-lived mode. CH4 responds via the primary tropospheric mode on a longer timescale (∼12 years) causing further O3 changes on this longer timescale. The radiative forcing from the short-lived O3 mode is regional, while forcing from CH4 and O3 due to the long-lived primary mode is relatively homogenous. The climate response to regional forcings will likely differ from the response to a well-mixed greenhouse gas that has a more globally uniform forcing. These basic differences in the temporal and spatial behavior of O3 and CH4 forcings complicate our estimate of the net climate forcing from perturbed regional emissions of O3 precursors.
 The magnitude of the net radiative forcing from changes in O3 and CH4 simulated for a 10% reduction in regional O3 precursor emissions is small (the net forcing for the emission reductions we examined for each region is three orders of magnitude smaller than the global present-day radiative forcing due to O3 of 0.35 Wm−2 [Ramaswamy et al., 2001]). However, we are only examining the marginal effect of a small reduction (10%) in the anthropogenic emissions from individual continents – each reduction considered is less than 2% of the total global anthropogenic emissions of O3 precursors. This unsurprisingly results in small perturbations in O3 and CH4 burdens and forcings. For comparison, reductions in aggregate anthropogenic CO2 equivalent emissions of the long-lived greenhouse gases agreed under the Kyoto Protocol are only 5% below 1990 levels for developed countries resulting in a relatively small decrease in radiative forcing. For a 10% reduction in European emissions of CO2 and air pollutants, Berntsen et al. [2005b] calculated the net climate forcing due to O3 reductions to be approximately an order of magnitude smaller than that due to CO2 reductions.
 In this study, we quantified the global change in tropospheric O3 and CH4 burdens and the associated radiative forcing resulting from regional emissions of O3 precursors (NOx, CO, and NMHCs). We evaluated the response of O3 and CH4 to reduced anthropogenic NOx emissions alone for nine geographical regions individually and to a combined reduction in anthropogenic emissions of NOx, CO and NMHCs from three regions, using chemical transport model simulations. We show that O3 and CH4 forcings are most sensitive to emission changes from tropical regions (Southeast Asia) and least sensitive to emission changes from midlatitude and high-latitude regions (Europe and North America). The range of normalized forcings found in our analysis suggests that control strategies that reduce emissions of O3 precursors from tropical regions in particular can have a significant impact on the net climate forcing from O3 and CH4. We find that from all regions of the world, reductions in NOx emissions alone result in a positive forcing from increased CH4 that dominates over the small negative forcing from decreased O3 for each region, implying a net warming. Combined reductions in anthropogenic NOx, CO, and NMHC emissions result in a stronger negative forcing from decreased O3 and a weaker positive forcing from CH4, producing a net negative forcing (cooling).
 The analysis presented here may prove useful in incorporating tropospheric O3 and its precursors in a future climate treaty to gain climate change benefits. Since we find that NOx emission reductions alone are insufficient to produce a net negative radiative forcing, it may therefore be useful to assess the cost effectiveness and political feasibility of including the basket of O3 precursors in a future climate agreement. Consideration of a climate treaty seeking to obtain cobenefits from reducing radiative forcing and mitigating air pollution could benefit from further studies pertinent to crediting simultaneous reductions of regional emissions of NOx, CO, and NMHCs, or possibly from CO and NMHCs alone.
 V. Naik was supported by the Carbon Mitigation Initiative (CMI) of the Princeton Environmental Institute at Princeton University (http://www.princeton.edu/∼cmi) which is sponsored by BP and Ford. We thank Michael Prather for helpful discussions and Jan Fuglestvedt, Terje Berntsen and an anonymous reviewer for valuable comments on an earlier version of the manuscript. The Geophysical Fluid Dynamics Laboratory provided necessary computational resources.