Global climate impact of civil aviation for standard and desulfurized jet fuel



[1] Aircraft emissions can affect climate change through increasing carbon dioxide (CO2) but also via a host of other short-lived non-CO2 effects that are complex, involve impacts that are both warming and cooling and are unique to this sector. Previous assessments of aviation climate impacts have used a segmented approach whereby each effect was calculated separately and the effects summed. Integrated approaches using newly available Earth System models that allow simulation of more realistic interactions between effects are now possible. The NASA GISS Earth System Model (ModelE) is applied to reassess the net climate impact of civil aviation emissions based on a new inventory for year 2006 developed using the Federal Aviation Administration's (FAA) Aviation Environmental Design Tool (AEDT). The model simulates all known non-CO2 aviation climate impacts except linear contrails and contrail-cirrus for which a recent estimate is assumed. For standard jet fuel, the net global climate impact for sustained constant year 2006 aviation emissions is +44 ± 10 mWm−2 (2/3 due to non-CO2 effects) on a 20-year timescale and +73 ± 10 mWm−2 (over 1/3 due to non-CO2 effects) on a 100-year timescale. For desulfurized jet fuel, the net climate impact is +40 ± 10 mWm−2 on the 20-year timescale, slightly less warming than the standard fuel case due to the complex interplay between sulfate and nitrate and the competition for ammonia. Ozone (O3) greenhouse efficiency (W per g O3 change) is 20–60% larger for aviation than surface transportation emissions.

1. Introduction

[2] Air transportation is fundamental in the globalized economy and is the fastest growing fossil fuel based sector with emissions projected to double over year 2005 values as soon as 2025. The industry challenge is to achieve capacity growth and mobility with reduced environmental impacts and enhanced energy efficiency and security. This goal requires improved understanding of the effects of aviation emissions on the global climate. Aviation is at present responsible for about 3% of all fossil fuel CO2 emissions, but an estimated 2–14% of anthropogenic climate forcing when all of the non-CO2 effects are considered [Lee et al., 2009]. Desulfurizing jet fuel is a mitigation strategy primarily motivated by a need to improve air quality around airports but the impact of this fuel switch on climate needs to be assessed.

[3] Aviation is a unique anthropogenic influence because the emissions are deposited directly at cruise altitudes (8–12 km) into the upper troposphere and lower stratosphere (UT/LS) and have considerable potential to influence the radiative balance of the Earth system in complex ways. Aircraft engines of the civilian fleet emit CO2 and water vapor (H2O), but also other chemical species that can influence climate, the most important of which are nitrogen oxides (NOx), sulfur dioxide (SO2) and black carbon (soot) particles. In the UT, NOx emissions participate in chemical reactions that tend to generate ozone (O3) resulting in warming, and increase the hydroxyl radical (OH), the major atmospheric oxidizing agent. The OH enhancements cause a reduction in the atmospheric lifetime of methane (CH4). This indirect CH4 change results in cooling and affects O3 on the longer timescale of the CH4 lifetime [Wild et al., 2001]. SO2 emissions are oxidized in the atmosphere to form sulfate aerosol particles. NOx emissions also contribute to nitrate aerosol formation and, through their effect on OH, can increase the oxidation of SO2 to sulfate [Unger et al., 2006]. Most aerosol particles scatter solar radiation back to space and lead to net cooling, except for black carbon, which absorbs solar radiation and warms the planet. There are additional effects on climate through changes in cloudiness in the upper troposphere via several pathways. Linear contrails and contrail-cirrus cloud from spreading contrails are both believed to cause warming since the long-wave trapping by high cirrus clouds dominates over the short-wave reflection with current estimates based on observed trends in cirrus cloudiness in the range +10–80 mWm−2 [Lee et al., 2010; Stordal et al., 2005; Stuber et al., 2006]. Aerosols emitted by aircraft and formed within the plume may act as ice nuclei and may form additional soot-cirrus or modify large-scale cirrus properties. The first global-scale estimates of the influence of aircraft soot on large-scale cirrus that applied the same basic modeling system (but some different assumptions and parameterizations) found differences in the sign of the impact ranging from −160 to −120 mWm−2 [Penner et al., 2009] and −110 to +260 mWm−2 [Liu et al., 2009]. These estimates would dominate the total aviation climate signal and may imply a net global cooling even on long time scales analogous to shipping emissions [Fuglestvedt et al., 2009]. In this study, linear contrails and contrail-cirrus are considered but not the interactions between aviation aerosol and large-scale cirrus because the fundamental mechanisms that drive these interactions are not yet understood.

[4] Landmark work applied a one-dimensional radiative convective-equilibrium model to show that O3 is 10 times more efficient as a greenhouse gas in the upper troposphere than in the lower troposphere mainly because of the thermal structure [Lacis et al., 1990]. Furthermore, aviation NOx emissions have a larger O3 chemical production efficiency than surface emissions on a per molecule basis because they occur at high altitudes where the NOx lifetime is relatively longer [Myhre et al., 2011]. Most previous assessments have determined the aviation climate effects in a fragmented approach, for example, separating O3 and aerosol responses [Balkanski et al., 2010; Hoor et al., 2009; Lee et al., 2010; Myhre et al., 2011] except for 2 studies that found dramatically reduced aviation O3 radiative forcing when heterogeneous chemistry on sulfate aerosol in the aircraft exhaust was taken into account [Pitari et al., 2002; Sovde et al., 2007].

[5] Here, a comprehensive reassessment of aviation impact on global climate is provided using the exact version of ModelE prepared for the forthcoming IPCC AR5 and a new hourly resolution inventory for year 2006 developed at the Volpe National Transportation Systems Center using the Federal Aviation Administration's (FAA) Aviation Environmental Design Tool (AEDT) [Wilkerson et al., 2010]. I follow the reporting structure applied in the European Commission Sixth Framework QUANTIFY project [Lee et al., 2009] based on an original IPCC assessment and updates [Sausen et al., 2005]. The radiative forcing (RF) concept provides a measure of the change in planetary energy caused by the addition of a particular agent in the Earth-atmosphere system and is widely used to quantify climate impact because on a global mean annual average basis RF is directly related to the global mean equilibrium surface temperature response via climate sensitivity. Here, I report instantaneous global mean annual average RF at the tropopause for each effect and for sustained constant emissions relative to the aviation free atmosphere. The impacts of aviation emissions operate on a wide range of time scales: minutes to hours (linear-contrails and contrail-cirrus); days to weeks (O3 and aerosols); about a decade (CH4); decades to centuries (CO2). Therefore, in the case of sustained constant emissions, the RF is time-independent for the short-lived effects, reaches steady-state after about 10–20 years for CH4 and increases over time for CO2. In order to encompass assessment of these different mechanisms, the net RF values are presented for 2 different timescales, the near-term 20-year that emphasizes the non-CO2 short-lived RF effects and the longer-term 100-year for which the effects of CO2 become more important. The relative importance of the non-CO2 impacts declines as the timescale increases. The RF calculations do not include stratospheric temperature adjustment that may reduce the O3 RF by 10–20% but have no significant affect on the aerosol RFs [Hansen et al., 2005].

2. Methodology

[6] Aviation emissions of NOx, SO2, black carbon, organic carbon, non-methane hydrocarbons and H2O are implemented into the NASA Goddard Institute for Space Studies (GISS) Earth System Model (ModelE) [Shindell et al., 2006] at hourly resolution (Table S1 and Figure S1 in the auxiliary material). About 14% of the aviation source is deposited directly into stratosphere on the annual average. The AR5 version of ModelE has horizontal resolution 2° × 2.5° latitude by longitude and 40 vertical layers. The model includes embedded atmospheric composition, interactive stratospheric and tropospheric chemistry and advanced treatment of gas-aerosol coupling in the exhaust and background atmosphere. The increased vertical resolution relative to the AR4 version has substantially improved the realism of the stratospheric circulation and the stratospheric chemical fields. Several updates have been made to the chemistry code relevant to aviation. For example: acetone has been added to the hydrocarbons included in the model [Houweling et al., 1998]; polar stratospheric cloud formation is now dependent upon the abundance of HNO3; H2O and temperature [Hanson and Mauersberger, 1988]; a reaction pathway for HO2 + NO to yield HNO3 has been added [Butkovskaya et al., 2007]; the coefficient for N2O5 uptake on aerosol surface is now dependent on temperature and humidity; and on-line aerosol tracers influence photolysis rates in FASTJ2. The aerosols are assumed to be externally mixed [Koch et al., 2009].

[7] ModelE is applied to determine aviation emission effects on O3, CH4 lifetime, sulfate, nitrate, black carbon, organic carbon, and H2O. Background anthropogenic emissions sources are from the IPCC Representative Concentration Pathway 4.5 for year 2005 [Wise et al., 2009]. CH4 concentration is prescribed to hemispherically averaged values (NH = 1814 ppb and SH = 1733 ppb) based on observations for the year 2005 [Dlugokencky et al., 2011]. Monthly mean sea surface temperatures and sea ice climatologies are prescribed for 1996–2005 [Rayner et al., 2006]. In this model version, aerosol effects on large-scale cirrus have not been included, and aviation aerosols have insignificant effects on liquid phase clouds. Linear contrails and contrail-cirrus are not simulated interactively in this model version so I apply a recent estimate of global RF by linear contrails and contrail-cirrus that has been determined in a process-based contrail-cirrus module in the ECHAM4 global climate model that accounts for the decrease in natural cirrus caused by the aviation effects (+31 mWm−2 ± 25%) [Burkhardt and Kärcher, 2011].

[8] Two core simulations were performed (with and without aviation emissions) based on the model configuration and emissions inventory described above and in Table S1. A simulation was performed for the desulfurized fuel case decreasing the fuel sulfur content from 600mgS/kg fuel to 15mgS/kg fuel. To investigate the altitudinal impact on the climate effects a simulation was performed that collapsed the aviation emissions into model level 1 but retained the horizontal spatial distribution. Further pairs of simulations (with and without aviation emissions) were performed for several sensitivity tests including: (1) switch off aviation SO2 and direct sulfate emission (2) switch off HNO3 uptake on mineral dust (3) switch off coupled nitrate aerosol module (4) switch off branching channel HO2 + NO → HNO3. For all the simulations, the induced changes in atmospheric composition are not allowed to feed back to the dynamics through the radiation, even though the model radiation scheme is used to determine the resultant RF. In this way, the dynamics are fixed between the runs and it is possible to obtain statistically significant signals without running for a large number of years. Thus, each simulation was run for 12 model years; the first 2 years of the simulations are discarded as spin-up and the remaining 10 years are averaged for analyses. The contributions to RF are then determined by taking the difference between the control simulation and the relevant sensitivity simulation. In each individual simulation, the radiation code is called twice for each component (with and without the component) at every 30-minute model time step. Aviation CO2 RF is calculated using the gas-cycle component of the Model for the Assessment of Greenhouse-gas Induced Climate Change [Wigley, 1994]. The aviation-induced impact on CH4 concentration is calculated based on the aviation-induced change in the CH4 lifetime in the climate model and accounting for the feedback of CH4 on its own lifetime following [Berntsen et al., 2005]. Then the CH4 RF is calculated using a simplified expression based on the steady-state concentration change and the secondary O3 RF is determined using global average sensitivity results from previous multi-model assessment studies [Berntsen et al., 2005].

3. Results

[9] On the annual and zonal mean, aviation emissions increase O3 concentrations by about 2–3% through the troposphere and do lead to small decreases (1–2%) in the polar LS (Figure 1) regardless of the fuel type. For standard fuel, aviation increases sulfate by up to 10–15% in the UT in the vicinity of the major flight tracks. Sulfate and nitrate aerosol formation are intimately coupled in the atmosphere due to their competition for available ammonia [Bauer et al., 2007; Unger et al., 2010]. In ModelE, the aviation-induced ammonium sulfate aerosol formed at high northern altitudes and latitudes is in part at the expense of ammonium nitrate aerosol that is not formed in the aviation-influenced atmosphere (decrease of 10–20%; Figure 1). At lower latitudes and altitudes over Asia, where the regional ammonia source is more abundant, aviation emissions result in a net production of both sulfate and nitrate aerosol (5% increases). For desulfurized fuel, the aviation-induced sulfate increases are much smaller, only about 2% through the troposphere. In this case, since the competition for ammonia is not imposed to the same extent as for the standard fuel, nitrate reductions over the polar region are only a few percent.

Figure 1.

Impact of aviation emissions on annual mean and zonal mean concentrations of ozone, sulfate and nitrate (%) for standard and desulfurized jet fuel.

[10] For the standard fuel case, the largest contributions to climate impact on short timescales are from contrail-cirrus, CO2, and CH4 (Figure 2). ModelE aviation O3 RF is less than the previous assessments: +6 mWm−2 (or 0.8 × 10−11 Wm−2/kgNyr−1) versus +13–26 mWm−2 (or 1.6–2.6 × 10−11 Wm−2/kgNyr−1) [Lee et al., 2009; Myhre et al., 2011] because the net O3 chemical production efficiency (molecule O3 per molecule aviation NOx emitted) is only about 1 compared to 1.5 to 2.4 for other models [Myhre et al., 2011]. A comparable O3 RF of 1.0 × 10−11 Wm−2/kgNyr−1 was obtained in an older AR4 version of ModelE that has coarser resolution (4° × 5° latitude by longitude and 23 vertical layers) based on the year 2005 emission inventory [Lee et al., 2009]. The O3 RF model sensitivity to the entire anthropogenic emissions change across the industrial era is in good agreement with IPCC best estimates [Shindell et al., 2006] even when the contributions of individual emission sectors are summed [Unger et al., 2010]. The aviation CH4 RF is similar to previous results. Hence, there is a net negative RF (cooling) from the combined O3 and CH4 response to aviation NOx emissions. A near balancing of the effects has been found in the same model for the shipping and industry sectors [Unger et al., 2010].

Figure 2.

Comparison of aviation RF impact for sustained constant emissions on 20-year timescale for the standard, desulfurized and surface emission cases. (top) Individual contributions and (bottom) the net sum. Error bars include uncertainty in aviation emissions and standard error relative to internal climate model variability based on 10 years of output. Uncertainty in contrail-cirrus RF is ±25% [Burkhardt and Kärcher, 2011].

[11] The ∼100% spread in existing aviation O3 RF assessments has been partially attributed to inter model variability in background NOx [Holmes et al., 2011; Stevenson and Derwent, 2009]. ModelE demonstrates skillful ability to simulate regional variability in NOx vertical profiles (Figure S2). It does appear that AR5 ModelE UT NOx may be too high compared to observations in regions around the North Atlantic relevant for aviation impacts although aviation NOx emissions play an unimportant role in these regional profile magnitudes. 60% of model NOx is within a factor of 2 of observed regional NOx climatologies (Figure S3). The UT high bias may be related to too vigorous deep convection in this model version. Future work will assess the effects of a modified convection scheme that increases the entrainment and makes less deep and more shallow convection.

[12] Despite the lower chemical production efficiency, ModelE O3 greenhouse efficiency (W per g O3 change) is strikingly similar to previous results for aviation and surface transportation modes (Figure S4). Greenhouse efficiency differences between transportation modes are driven by differences in the spatial location of the induced O3 perturbation. A linear fit to each sector gives O3 greenhouse efficiencies in W/g (n = number of different model assessments): aviation = 1.8 (R2 = 0.73, n = 15), shipping = 1.1 (R2 = 0.76, n = 11); road vehicles = 1.5(R2 = 0.74, n = 12). Thus, O3 greenhouse efficiency from aviation emissions is about 20–60% greater than for surface-based transportation emissions. This sensitivity is much less than a factor of 10 [Lacis et al., 1990] because the tropospheric lifetime of O3 is longer than the timescale for vertical mixing. The induced O3 perturbation is fairly well mixed throughout the tropospheric column regardless of the location of the initial precursor injection and this result appears robust across the different modeling systems in Figure S4. A corollary is that O3 precursors emitted at the surface are of greater climate concern than previously assumed.

[13] Effects of aviation aerosols via heterogeneous chemistry do not play a significant role in the ModelE aviation O3 response consistent with previous work that found only weak aerosol feedbacks on gas-phase chemistry [Shindell et al., 2009] but inconsistent with other models [Pitari et al., 2002; Sovde et al., 2007]. Complete removal of aviation SO2 (and therefore sulfate) has only a small impact on the aviation O3 response (∼7% decrease). Similarly, turning off heterogeneous uptake of nitric acid (HNO3) on dust and the interactive nitrate aerosol simulation has only minimal impacts on the aviation O3 response. The ModelE chemical mechanism does include the HNO3 forming branch of the HO2 + NO reaction [Butkovskaya et al., 2007]. Accounting for this new branch may result in a 20% reduction in the preindustrial to present day O3 RF [Sövde et al., 2011]. In ModelE, eliminating this branch from the chemical scheme has negligible impact on the aviation O3 response.

[14] The magnitude of the aviation sulfate RF is larger than previous estimates: −2.8 × 10−11 versus −0.6 to −2.6 × 10−11 Wm−2/kgSO2yr−1 [Balkanski et al., 2010; Lee et al., 2009] because here the model accounts for additional oxidation of non-aviation SO2 to sulfate via aviation NOx effects on atmospheric oxidation [Unger et al., 2006]. The aviation-influenced atmosphere drives a net reduction in nitrate aerosol loading resulting in a positive RF that largely balances the sulfate cooling.

[15] For standard jet fuel, the total non-CO2 RF (+29 mWm−2) is approximately double the CO2 RF (+15 mWm−2) giving a net aviation climate impact on a 20-year timescale of +44 ± 10 mWm−2. On a 100-year timescale the CO2 RF is +44 mWm−2 giving a net climate impact of +73 ± 11 mWm−2. The altitudinal importance of the source impact is demonstrated by collapsing the aviation emissions into the surface model layer, which results in a halving of the O3 and CH4 RF (Figure 2). In this case, there is no impact on contrail-cirrus; nitrate becomes the dominant aerosol effect whereas the sulfate and black carbon responses are tiny. The combined non-CO2 is a cooling (−6 mWm−2) that offsets the CO2 warming by 40% yielding a net impact of +9 ± 3 mWm−2. Global-scale desulfurization of jet fuel has only small impacts on the O3 and CH4 aviation RF (Figure 2). For this fuel type, the sulfate RF is substantially reduced relative to the standard fuel and the nitrate RF changes sign to a small net cooling effect. Because of the interaction between the inorganic aerosol components, the total non-CO2 RF from desulfurized fuel is less positive than standard fuel (+24 mWm−2). Therefore, the net climate impact is slightly less warming (+40 ± 10 mWm−2), counterintuitive on the basis of the anticipated warming uplift due to sulfate removal. Future work will investigate sensitivity of this result to the ammonia distribution in the UT, which has received relatively little attention.

4. Conclusions

[16] To date, aviation emissions have not been included in any cross-sector emissions trading scheme but will be included in the European Union scheme in January 2012. Accounting for non-CO2 impacts in cross-sector emissions trading schemes is controversial, but some have proposed using a simple multiplicative factor between 2 and 4 to be applied to aviation CO2 emissions. I conclude that enough uncertainty still exists around understanding of the non-CO2 aviation impacts to prohibit their inclusion in emission trading in agreement with previous work [Forster et al., 2007]. For climate protection, the reassessment provided here does not support application of a NOx stringency policy option in which modest reductions in aviation NOx emissions are traded against modest increases in CO2 but does suggest that desulfurizing jet fuel may have a slightly less warming impact than standard fuel, thus air quality goals can be achieved without increasing global climate warming. The robustness of this finding across other global modeling systems needs to be investigated. There are further potential indirect climate effects of fuel desulfurization through interactions between aerosol and large-scale cirrus, which may be important, but meaningful quantitative estimates cannot be obtained due to large uncertainties in the fundamental mechanisms and even the sign of the effects.

5. Additional Information

[17] 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 author and do not necessarily reflect the views of the US DOT Volpe Center, the US FAA, EUROCONTROL or ICAO.


[18] Support for this research is provided by the Aviation Climate Change Research Initiative (ACCRI) under contract DTRT57-10-C-10013. The author gratefully thanks: D. Rind and A. Del Genio for helpful discussions; R. Halthore, D. Jacob and M. Gupta for program guidance; and M. Kelley for assistance with the emissions data. Computational support was provided in part by the facilities and staff of the Yale University Faculty of Arts and Sciences High Performance Computing Center and by the National Science Foundation under grant CNS 08-21132 that partially funded acquisition of the facilities. The author wishes to thank Joyce Penner and an anonymous reviewer for their assistance evaluating this paper.

[19] The Editor wishes to thank Joyce Penner and an anonymous reviewer for their assistance evaluating this paper.