Strong growth in the civil aviation sector will accelerate in the future. Here, we confront the future net chemical (ozone, methane, sulfate, nitrate, black carbon, and water vapor) global climate impact of aviation at 2050 for three novel plausible scenarios constructed at the Volpe National Transportation Center using the U.S. Federal Aviation Administration (FAA) Aviation Environmental Design Tool (AEDT). The aviation net chemical climate impact is cooling in all cases and increases from −10 ± 4 mW m−2 in the contemporary climate up to –69 ± 21 mW m−2 by 2050. Future improvements in fuel efficiency provide the opportunity to reduce aviation's net chemical climate impact by ~50% relative to a baseline scenario of unconstrained growth. On the 20 year time horizon, the cooling net aviation chemical climate impact masks the aviation CO2 global warming by up to 50–100% in the contemporary and future atmospheres.
 Civil aviation is the fastest growing fossil fuel-based sector today. The strong growth is expected to accelerate in the future as air travel becomes more widespread and economically accessible to the growing middle class in the developing world. Assessment of the future impacts of this growth on global climate is essential for effective environmental management of the aviation industry and the development of appropriate mitigation strategies. [Lee et al., 2009]. Aviation is at present responsible for less than 2% of all anthropogenic global annual carbon dioxide (CO2) emissions, yet a substantially larger (2–14%) fraction of anthropogenic climate impact due to significant augmentation by a range of other non-CO2 effects [Lee et al., 2010]. The most recognized non-CO2 climate effects are the following: NOx emissions that increase tropospheric ozone (O3) (warming) and decrease the lifetime of methane (CH4) (cooling); SO2 emissions that increase sulfate aerosol (cooling); soot or black carbon (BC) emissions (warming); formation of persistent linear contrails and induced cirrus (warming); and water vapor (H2O) emissions (warming). An aviation-induced nitrate aerosol response was recently discovered [Barrett et al., 2012; Unger, 2011].
 The climate impacts of aviation are quantified using the radiative forcing (RF) concept, which provides a measure of the change in planetary energy caused by the addition of a particular agent in the Earth-atmosphere system. On a global mean annual average basis, RF is directly related to the global mean equilibrium surface temperature response via climate sensitivity. We define the net chemical climate impact from aviation as the sum of the short-lived non-CO2 gas and direct aerosol RFs (O3, CH4, sulfate, nitrate, black carbon, and H2O). The O3, CH4, H2O vapor, sulfate, and nitrate effects are strongly coupled through gas-phase photo-oxidation chemistry that controls the lifetime of CH4 and the formation of the secondary inorganic aerosols and O3. Persistent linear contrails and induced cirrus may be the largest single climate impact of aviation emissions [Burkhardt and Karcher, 2011], yet the net chemical RF, which includes both positive and negative component values, rivals the contrail-cirrus RF [Lee et al., 2010]. Previous assessments of future aviation chemical RF have relied on a simplified linear approach whereby aviation radiative efficiencies of the non-CO2 components (RF per unit precursor emission) derived from complex simulations of the contemporary world are multiplied by projected emission increases. This simplified scaling to obtain the chemical RFs due to future aviation emission increases does not account for the influence of changing background conditions, chemical interactions between short-lived components, and the sensitivity of the non-CO2 RFs to the spatial location of emissions [Berntsen et al., 2005; Rypdal et al., 2005; Stevenson and Derwent, 2009]. A newly developed set of scenarios, the Representative Concentration Pathways (RCPs), is currently in use for ongoing international, multimodel activities in support of the upcoming Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) [van Vuuren et al., 2011]. All RCPs assume aggressive air pollution abatement measures and correspondingly large decreases in aerosol and O3 precursors globally. Concurrently, the Volpe National Transportation Systems Center has applied the U.S. Federal Aviation Administration (FAA) Aviation Environmental Design Tool (AEDT) to develop three alternative high-resolution future 2050 projections of aviation emissions based on a broad range of industry plausible scenarios.
 Here, we apply the new generation NASA Goddard Institute for Space Studies general circulation model (ModelE2) global coupled oxidant-aerosol-climate model to compute the future aviation global chemical RFs based on the Volpe scenarios. The future net chemical climate impact is examined within the context of the future aviation CO2 climate impact. We select the midrange RCP4.5 as the background world for all the aviation scenarios. RCP4.5 is a CO2 stabilization pathway without overshoot to 4.5Wm−2 (~ 650 ppm CO2 eq) after 2100. Historical, transient aviation emissions between 1940 and 2050 are not available for the Volpe scenarios. Thus, our metric of choice is the global mean annual average RF on the 20 year and 100 year time horizons based on sustained constant emission rate in the contemporary and future time slices. This metric is partly analogous to the global warming potential and fully consistent with our previous assessments [Unger, 2011; Unger et al., 2010].
 Table S1 in the auxiliary material presents the Volpe global (nonmilitary) aviation emission inventory estimates for the year 2006 (2006-BASE) and three future 2050 cases: the Baseline Scenario (2050-BASE), the Technology and Operational Improvements Scenario (2050-S1), and the Renewable Alternative Fuels Scenario (2050-S2).1 2050-BASE describes unconstrained growth with a factor of 4.8 increase in fuel burn over year 2006. 2050-S1 is based on the International Civil Aviation Organization's aspirational fuel efficiency goal of 2% per annum to 2050 with a factor of 2.7 increase in fuel burn over year 2006 and includes additional NOx emission improvements. 2050-S2 is an alternative scenario focused on air quality improvements that builds on 2050-S1 with the additional assumption of the complete penetration of fully formulated sustainable alternative fuels such that fuel composition in 2050 is sulfur-free with a maximum aromatic content of 8%. CO2 emissions are not available for 2050-S2, and so here we adopt the 2050-S1 CO2 emission values for this scenario. The Volpe 2050 projections do not closely resemble any of the RCPs' own internal aviation scenarios. For instance, the global aviation source of NOx and CO2 in 2050-BASE is double that in RCP8.5, and while 2050-S1 and 2050-S2 have similar aviation NOx to the RCP4.5, CO2 is 50% higher.
 We pair the year 2006 Volpe aviation emissions with RCP4.5 for year 2005 (as broadly representative of anthropogenic emissions in the contemporary world) and year 2050 Volpe scenarios with RCP4.5 at 2050. RCP4.5 precursor emissions at year 2005 and year 2050 from all other non-aviation anthropogenic sectors are detailed in Table S2. Substituting the Volpe aviation emission totals into the RCP4.5 inventory gives relatively small aviation sector fractional contributions to total anthropogenic sources for present-day (2006-BASE): CO2 = 1.8%, NOx = 2.1%, and BC = 0.07%. The aviation sector contributes much larger fractional sources in 2050-BASE: CO2 (6.6%), NOx (12.5%), and BC (0.5%), whereas 2050-S1 contributions are more modest: CO2 (3.9%), NOx (5.4%), and BC (0.3%). The spatial distribution of the present and future aviation NOx emission source is presented in Figure S1.
 The global gridded Volpe aviation emissions of NOx, CO, SO2, BC, organic carbon (OC), H2O, alkenes, and paraffins are implemented at hourly resolution into NASA ModelE2. This global chemistry-climate model (CCM) has been substantially improved for simulations in support of the IPCC AR5 [Schmidt et al., 2006]. The horizontal resolution is 2° latitude by 2.5° longitude with 40 vertical hybrid sigma pressure layers from the surface to 0.1 hPa. The tropospheric and stratospheric gas-phase chemistry and aerosol modules are fully integrated so that these components interact with each other and with the physics of the climate model [Bell et al., 2005; Koch et al., 2011; Shindell et al., 2012].
 We define four simulation scenarios named according to the Volpe aviation emission inventories in Table S1. Time-slice simulations are performed for the present-day and future worlds by prescribing 5 year average monthly-varying sea surface temperatures and sea ice boundary conditions extracted from previously completed, fully coupled, transient simulations of the RCP4.5 scenario using the exact same climate model (L. Nazarenko and R. Ruedy, personal communication). 2006–2010 (2048–2052) averages are applied for the present-day ~2005 (future ~2050) simulations. Greenhouse gases (GHGs) are prescribed to the RCP4.5 defined concentrations for year 2005 (CO2 = 379.3 ppmv, N2O = 319.4 ppbv, CH4: Northern Hemisphere (NH) = 1814 ppbv and Southern Hemisphere (SH) = 1733 ppbv) and year 2050 (CO2 = 486.5 ppmv, N2O = 350.6 ppbv, CH4: NH = 1925 ppbv and SH = 1741 ppbv). Annual mean global average surface air temperature and precipitation increase +1.82°C and 0.12 mm/day between the contemporary and future 2050 climates, respectively. The climate-sensitive global lightning NOx source increases modestly from 7.4 TgN/yr in 2005 and 7.8 TgN/yr in 2050.
 We perform two control simulations representative of the contemporary world and 2050, zeroing out the aviation emissions, and four sensitivity simulations with the aviation emissions inventories described in Table S1 and driven with the relevant RCP4.5 ocean boundary conditions and GHG concentrations for the present or future world. The simulations include the effects of feedbacks from future physical climate change. Integrations of 12 model years were completed; the first 2 years of the simulations are discarded as spin-up, and the remaining 10 years are averaged for analyses. The aviation signal is small compared to other precursor emission sources and climate model internal variability. In order to obtain statistically significant aviation perturbation RF signals without running for a larger number of years, we switch off the radiative feedbacks to dynamics from the in-line changes in atmospheric chemical composition.
 We isolate the contribution of aviation emissions to O3, sulfate, nitrate, BC, and H2O atmospheric composition and RF from the difference between the relevant sensitivity and control simulations. Aviation H2O RF was calculated in the model only for 2006-BASE and scaled to emission increases for the future scenarios. The aviation-induced impact on CH4 RF includes the secondary longer-term O3 RF response and is calculated off-line using the model's simulated CH4 chemical lifetime changes as detailed elsewhere [Unger et al., 2010]. The aviation CO2 RF for each time horizon is determined based on the IPCC absolute global warming potential [Forster, 2007].
 The absolute and fractional contributions of aviation emissions to annual zonal average O3, sulfate, and nitrate concentrations for 2006-BASE, 2050-BASE, and 2050-S1 are shown in Figures S2a and S2b. Results for 2006-BASE are similar to those reported previously [Unger, 2011] except for the nitrate aerosol response. In the previous work, aviation emissions in the contemporary climate contributed a 10–20% decrease in upper troposphere lower stratosphere (UTLS) annual zonal average nitrate concentrations and a 5% increase at lower latitudes and altitudes in the NH. Here, aviation contributes a similar amount to nitrate aerosol in the lower troposphere but only 2–3% decreases in the UTLS. Sulfate-nitrate-ammonium aerosols are formed in atmosphere through oxidation of the precursor gases (SO2 and NOx) and subsequent neutralization by available ammonia (NH3). Ammonium sulfate is the preferential species due to its lower vapor pressure. The higher volatility ammonium nitrate aerosol may be formed if excess free NH3 is available beyond the ammonium sulfate requirement. The effect of aviation SO2 and NOx emissions on the sulfate-nitrate-ammonium aerosol system depends on the availability of NH3 and the local meteorological and chemical environment. In NH3-limited environments such as the UTLS, additional SO2 emissions may lead to reductions in ammonium nitrate aerosol owing to ammonium sulfate formation at the expense of ammonium nitrate. The strong nonlinearity and acute temperature sensitivity of the sulfate-nitrate ammonium system is well established [e.g., Capps, 2012; Pye et al., 2009]. Figure S3 compares the aviation fractional contribution to nitrate aerosol and the background NH3 distributions in the previous work to this study. Our previous work used the aerosol development branch model version that had slightly different meta-version moist physics with a higher NH3 solubility coefficient than for the published value coded in the standard frozen IPCC AR5 model version that is applied in this work. Consequently, the model version in this study has higher NH3 concentrations throughout the troposphere by around a factor of 2 on the annual zonal average. The result is that the injection of aviation SO2 imposes less of a reduction in background nitrate aerosol in order to form aviation sulfate aerosol in the UTLS region. Understanding the effects of aviation on secondary inorganic aerosol formation and direct aerosol RF relies on the global NH3 simulation especially in the UTLS. Observational constraints for upper troposphere NH3 that are of use in global modeling are currently extremely limited to the extent that it is not possible to discriminate which model version provides the most accurate annual zonal average NH3 simulation. Newly available satellite NH3 observations from the Tropospheric Emission Spectrometer may hold some promise for better large-scale evaluation of NH3 distribution in the future [Shephard et al., 2011]. In 2050-BASE, aviation contributes up to 80–90% of the sulfate aerosol in the high northern latitude UTLS region, up to 40% of the nitrate aerosol in the midtroposphere at NH midlatitudes, and 5–7% of the O3 throughout the tropospheric column between 30–60°N. In 2050-S1, aviation fractional contributions to the trace composition are about half those in 2050-BASE.
 The present and future aviation global climate impacts for individual component, scenario, and time horizon are presented in Figure 1 and detailed in Table S3. The model's contemporary aviation O3 RF is less than in other model estimates because the O3 chemical production efficiency to aviation NOx is only about half that in other global models [e.g., Myhre et al., 2011]. Other published model estimates are based on the application of global chemistry-transport models (CTMs) that are driven with fixed meteorology from reanalysis datasets updated every 4–6 h, and that typically do not incorporate interactive stratospheric chemistry or a coupled hydrological cycle. It is not surprising that these modeling frameworks gave similar aviation O3 responses to each other, which may have led to the false impression of higher certainty and robustness of the aviation O3 impact than exists in reality. Intermodel differences in the O3 sensitivity to aviation NOx are mostly a result of the intermodel variability in the background NOx levels [Holmes etal., 2011; Stevenson and Derwent, 2009]. The major hindrance to reducing the uncertainty is that existing NOx measurements in the UTLS, amounting to a few sporadic regional aircraft campaigns, some of which are 20 years old or more, are not sufficient to characterize accurately the background NOx across the spatial and temporal scales necessary for evaluation of global CCMs and CTMs. A recent comprehensive evaluation of NASA ModelE2 identified a possible positive bias in near-tropopause NO2 in midlatitudes in both hemispheres against OMI NO2 column measurements [Shindell et al., 2012]. The cause was attributed to too rapid stratospheric circulation (and too little stratosphere-troposphere exchange especially in the SH). Such a bias would impact all NOx-related aviation effects. We find that a ModelE2 simulation nudged to large-scale meteorology from the GMAO MERRA reanalysis [Rienecker et al., 2011] averaged over years 2000–2005 yields an aviation O3 RF only 30% higher than that with the climate model's meteorology, suggesting that the model's dynamics are playing a relatively small role in the reduced sensitivity to aviation NOx. Similarly, we find that a 50% spatially homogeneous reduction in lightning NOx yields only a 30% increase in aviation O3 RF, suggesting that a discrepancy in the lightning source is also not likely to be responsible for the reduced sensitivity. The dearth of useful trace gas measurements in the UTLS limits our ability to make a conclusive appraisal of the aviation O3 RF at this stage. We posit that aviation O3 RF is less certain than previous assessments have claimed.
 The aviation nitrate RF for the contemporary climate is a small net positive value, +7 mW m−2, reflecting the global average difference between the larger regional production and loss terms. The future aviation O3, sulfate, and BC responses essentially scale linearly with their primary precursor emission increase for each scenario, whereas the future nitrate and CH4 RFs are markedly nonlinear in response to NOx emission changes because of concurrent changes in the RCP4.5 background emissions (Table S3). The superlinear nitrate aerosol response at 2050 to aviation NOx change (RF values for 2050-BASE and 2050-S1 are ~10 and ~4 times larger than 2006-BASE, respectively) is driven by simultaneous reductions in background SO2 emissions and sulfate aerosol that result in increases in free NH3 available for ammonium nitrate formation. The increased importance of nitrate aerosol in the future global atmosphere as sulfur is removed, but NH3 emissions remain constant or increase has already been highlighted [Bauer et al., 2007; Bellouin et al., 2011; Pye et al., 2009]. Aviation CH4 RF demonstrates a consistent sublinear response to future growth implying that aviation NOx is slightly less efficient at increasing hydroxyl radical within the context of the future RCP4.5 background precursor emissions.
 The net aviation chemical climate impact for each scenario (in mW m−2) is as follows: −10 ± 4 (2006-BASE); −69 ± 21 (2050-BASE); −31 ± 10 (2050-S1); and −20 ± 9 (2050-S2). Remarkably, the net aviation chemical climate impact is negative (cooling) for all scenarios, and the future net chemical climate impact ranges from a factor of 2–7 larger in absolute magnitude than the contemporary climate impact depending on scenario. The fuel efficiency and technology improvements in 2050-S1 result in a significant (~50%) reduction in the net aviation chemical climate impact relative to 2050-BASE. The additional complete penetration of renewable alternative fuels in 2050-S2 offers a further ~30% reduction in aviation net chemical climate impact over 2050-S1.
 The combined aviation global climate impact including the net chemical and CO2 effects for each scenario on the 20 year time horizon (in mW m−2) is as follows: +5 ± 4 (2006-BASE); −1 ± 24 (2050-BASE); +9 ± 11 (2050-S1); and +19 ± 11 (2050-S2), increase on the 100 year time horizon to the following: +42 ± 8 (2006-BASE), +177 ± 42 (2050-BASE), +110 ± 23 (2050-S1), and +120 ± 23 (2050-S2). Thus, for 2050-BASE, on the 20 year shorter time horizon, the net aviation chemical cooling climate impact entirely offsets the CO2 warming impact resulting in an overall global effect close to zero. For 2006-BASE, 2050-S1, and 2050-S2, the net aviation chemical climate cooling impact masks the CO2 20 year warming impact by 50–75% resulting in an overall small global positive (warming) effect on the shorter time horizon. For all scenarios on the longer 100 year time horizon, the net aviation chemical climate impacts mask the overall CO2 warming by 15–30%.
 The aviation sector climate impact has important local, regional, and hemispheric variability because of the short-lived non-CO2 chemical effects (Figure 1). The combined climate impact including the net chemical and CO2 effects is strongly hemispherically asymmetric on both short and long time horizons with a greater warming impact for aviation in the SH than the NH. Indeed, the combined net chemical and CO2 climate impact is cooling (negative RF) in the NH on the 20 year time horizon for all scenarios (except 2050-S2). The local combined RF can reach −40 mW m−2 (2006-BASE), −200 to −250 mW m−2 (2050-BASE), and −40 to −80 mW m−2 (2050-S1) in the North Atlantic flight corridor. Hence, the net positive contrail-cirrus RF must exceed these values to yield an overall local positive RF. [Burkhardt and Karcher, 2011] do indicate that in the present-day regional net RF by contrail-cirrus is larger than +100 mW m−2 over the U.S., Europe, and the North Atlantic flight corridor implying an overall small local positive RF for aviation.
 The aviation net chemical climate impact is globally cooling for present-day and all future scenarios. Simplified scaling of aviation radiative efficiency by future precursor emission change does seem to provide a useful, rapid, low-cost approximation to compute global aviation net climate impact for some non-CO2 components (implying insensitivity to the background emission scenario and conditions), but this approach is wholly inadequate for CH4 and nitrate aerosol. Reducing uncertainties in the aviation chemical climate impact necessitates better availability of NOx and NH3 measurements in the UTLS. Aviation climate impact studies have tended to focus on the global scale. We contend that more research emphasis is needed on the regional climate response to aviation emissions for which the impacts may be substantially more important than for the global scale [Jacobson, 2010].
 The authors thank two anonymous reviewers for valuable comments on the manuscript. Support for this research is provided by the Aviation Climate Change Research Initiative (ACCRI) under contract DTRT57-10-C-10013. Computational support is provided by the facilities and staff of the Yale University Faculty of Arts and Sciences High Performance Computing Center. The authors thank the NASA ModelE2 development team. The emissions inventories used for this work were provided by US DOT Volpe Center and are based on data provided by the US Federal Aviation Administration and EUROCONTROL in support of the objectives of the International Civil Aviation Organization Committee on Aviation Environmental Protection. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the US DOT Volpe Center, the US FAA, EUROCONTROL or ICAO.