4.2.3. Model Results for Subsonic Aircraft Emissions

4.2.3.1. Ozone Perturbation

Figure 4-1 presents annual zonal average increases of O3 volume mixing ratios caused by aircraft NOx emissions predicted by the six models for the year 2015. As this figure shows, the models treat the tropopause significantly differently, which leads to qualitatively different O3 distributions and calculated O3 perturbations near the tropopause. The UiO model calculates a maximum increase of O3 of about 9 ppbv around 40-80°N at an elevation of 10-13 km. Throughout most of the Northern Hemisphere, increases larger than 1 ppbv are calculated. The IMAGES/BISA and HARVARD models calculate somewhat smaller peak perturbations of about 7 ppbv. In contrast, the Tm³/KNMI and the UKMO models calculate maximum changes of about 11 ppbv. The UKMO model computes large increases up to the 16-km level, probably as a result of relatively large vertical exchange rates in the vicinity of the tropopause. In contrast to the other models, the ECHAm³/CHEM model predicts the highest O3 perturbations in the Northern Hemisphere and Southern Hemisphere LS. Tropospheric changes are smaller than in the other models, however. The difference may be partly a result of the length of the ECHAm³/CHEM model run. This model has a representation of the stratosphere and reported results from the average of the last 10 years of a 15-year simulation-long enough to propagate aircraft emissions and O3 perturbation in the Northern Hemisphere to the Southern Hemisphere (via the stratosphere). The other 3-D models do not account for such stratospheric transport processes because they constrain O3 concentrations in their upper model levels: They fix their upper model layer concentrations using observations, or they prescribe O3 fluxes from the stratosphere into the troposphere. The use of these boundary conditions could lead to a calculated impact on stratospheric O3 that is too small. It may also be that the ECHAm³/CHEM model has too efficient transport in the LS.

The effect of constraining concentrations and fluxes at the upper boundary of the 3-D models was checked by running the stratospheric 2-D Atmospheric and Environmental Research, Inc. (AER) model for the same subsonic scenario. Consistent with the 3-D models, the AER model calculates a maximum O3 increase of 8-10 ppbv in the Northern Hemisphere at an altitude of 8-12 km. In the stratosphere at 16 km, small increases of 2 ppbv in the Southern Hemisphere and 6 ppbv in the Northern Hemisphere are calculated-somewhat higher, but consistent with most 3-D models. Calculated O3 increases are strongest in the UT and the LS. In the lower troposphere (< 6 km), the increase is reduced by a factor of about 5 in mixing ratio compared to the UT. All models calculate that about 85% of the O3 increase for 1992 is in the Northern Hemisphere; for 2015 and 2050, the portions are about 80 and 75%, respectively.

Although emissions of precursor NOx are spatially distributed heterogeneously, the resulting O3 increases are distributed more uniformly as a result of the combined effects of strong longitudinal mixing and the relatively long residence time of O3 in the free troposphere and LS. All models show efficient transport of excess O3 from source regions at mid-latitudes to high latitudes, where the residence time of O3 is particularly long as a result of decreased deposition (Stevenson et al., 1997; Wauben et al., 1997; Berntsen and Isaksen, 1999).

There may be a strong seasonal cycle in the calculated impact of aircraft emissions on O3. For example, using the same emission scenarios, the UiO and the UKMO models calculate a 40% larger increase of O3 in the Northern Hemisphere in April compared to July (Stevenson et al., 1997; Berntsen and Isaksen, 1999). Other models find much weaker seasonal cycles (e.g., IMAGES/BISA and ECHAm³/CHEM), or find maximum increases in summer (e.g., Tm³/KNMI and HARVARD). These seasonal differences are probably associated with different background NOx conditions in the different models (see Section 4.2.3.2).

4.2.3.2. NOx Perturbation

Figure 4-2 shows calculated zonal average increases of NOx from aircraft emissions in July 2015. In the Northern Hemisphere, all but one model calculate increases of up to 150 pptv. These increases can be compared with background levels of 50-200 pptv at northern mid-latitudes in the 12-km region. In the stratosphere, the ECHAm³/CHEM model calculates larger increases, probably as a result of more efficient transport to the stratosphere. The height distribution of NOx increases is very similar among models. All models also predict noticeable increases in upper tropospheric NOx at low latitudes in the Southern Hemisphere. Only small increases are estimated in the lower troposphere. Background NOx conditions, however, are rather different in the 3-D CTMs. For instance, at 12 km at 50°N, calculated background NOx mixing ratios may vary, depending on the season, by a factor of 2 to 4. Such large differences could be important for O3 production because of the nonlinear dependency of net O3 production on NOx concentrations, although, for the model scenarios explored, the global O3 increase appears to be almost linear for most of the anticipated NOx injections in the models (see Figure 4-3).

4.2.3.3. Future Total Ozone Increases from Aircraft Emissions and Comparison with Increases from Other Sources

Figure 4-3 presents the increase of global O3 abundance up to 16 km from aircraft emissions. The annual emissions of 0.5, 1.27, and 2.17 Tg N correspond to (projected) emissions for 1992, 2015, and 2050, respectively. Calculated O3 increases range from 4 to 7 Tg in 1992, 9 to 17 Tg in 2015, and 19 to 24 Tg in 2050. For each CTM, the global O3 increase scales almost linearly with aircraft NOx emissions-even for the 2050 high-demand sensitivity study G (3.46 Tg N yr^-1). However, O3 production is less efficient at high NOx emissions. The nearly linear response of global O3 increase to aircraft NOx emissions was not anticipated, given the well-known nonlinear O3 production as a function of NOx (discussed in Chapter 2). The main explanation seems to be that aircraft NOx and associated reservoir species (e.g., HNO3) are transported out of aircraft corridors, where net O3 production depends more linearly on NOx. We should recognize, however, that all of the global models used in this study have a coarse resolution that may systematically overestimate the O3 production. Secondary effects are background increases in CH4 and CO (as a result of enhanced surface emissions in 2015 and 2050), leading to somewhat more efficient O3 production per NOx molecule emitted, and the shift of emissions from Northern Hemisphere mid-latitudes toward the tropics, where background NOx concentrations are smaller.

To further test linearity in O3 increases to NOx emission beyond the upper limit selected in these model studies, an extremely high NOx emission of 1.5 times the high-demand, low-technology case (Scenario G) was run with the UiO model. This scenario showed only slight nonlinearity at lower NOx emissions. This extreme simulation indicated that a level of nonlinearity was reached at northern latitudes where O3 increases of only 10% were obtained, whereas at southern latitudes, where emissions are smaller, O3 increases (approximately 50%) were nearly linear with increases in NOx emissions.

Figure 4-4 shows the global increases of total O3 from aircraft emissions in 2015 and 2050 relative to those in 1992-that is, the difference of O3 budgets for scenarios listed in Table 4-4 (D with respect to B and F with respect to B). The same figure also shows the increases of O3 in 2015 and 2050 from the effects of changes in surface emissions (Section 4.2.2.1)-that is, the differences for scenarios C with respect to A and E with respect to A. Aircraft emissions account for approximately 15-30% of the total O3 increase in 2015 and 15-20% in 2050. It should be noted, however, that the projections of aircraft emissions and the IPCC IS92a scenario underlying the increase of surface emissions are extremely uncertain. Changes in aircraft or surface emissions scenarios could change the relative contribution from aircraft emissions to O3 perturbations significantly. Using scenario G (high demand), an approximately 45% higher increase of O3 from aircraft is calculated in 2050 by the UiO model (see sensitivity studies).

4.2.3.4. Influence of Changing OH on CH4 Lifetime

The reduction in CH4 lifetime would lead to a nearly uniform CH4 reduction globally because of the relatively long residence time computed for CH4. This result would be in contrast to O3, for which changes would occur on large regional scales. Finally, it should be noted that for computational reasons, the experiments in this assessment were performed using fixed CH4 concentrations at the Earth's surface (see Section 4.2.2.1), and CH4 concentrations at the surface were not allowed to adapt to higher OH abundances (positive feedback). Hence, even larger CH4 decreases are to be expected if CH4 ground flux boundary conditions are used. However, such calculations are much more computationally intensive. IPCC (1995) and Fuglestvedt et al. (1996) showed that the feedback factor is uncertain and model-dependent but is estimated to be in the range 1.2-1.5. Adopting a factor of 1.4 increases the percentage changes, in the CH4 lifetimes shown in Table 4-5, to -2.2 to -4.1% in 2015 and -3.2 to -6.0% in 2050. Changes in CH4 lifetime of this order will lead to global average radiative forcing (see Chapter 6) similar to global average radiative forcing perturbations from aircraft-induced O3 changes, but with an opposite sign (CH4 will be reduced).

4.2.3.5. Sensitivity Studies

In this section, we focus on a limited set of sensitivity studies that help define the uncertainty range of the model calculations. Ideally, a large number of simulations should have been performed by all the participating models. However, only a limited number of model simulations was possible because of time constraints and the demand of computer resources for 3-D CTM studies. Therefore, only one or two models have performed each of the sensitivity studies. Also, some uncertainties cannot be addressed properly. For example, modeled upper tropospheric and lower stratospheric NOx and NOy concentrations are extremely uncertain and are difficult to compare to measurements because of large temporal and spatial variations and a limited number of observations (see Chapter 2). There may be additional uncertainties from unknown processes that feed back on increases in NOx and O3 in a future modified atmosphere.

The sensitivity studies focus on the impact on O3. The following studies were performed:

Sensitivity of aircraft-induced O3 perturbations to background NOx levels from lightning. This finding is obtained by increasing global average NOx production from 5 Tg N yr^-1, which is used in the reference case, to 12 Tg N yr^-1. The same spatial distribution is used in both cases.
Sensitivity of O3 perturbation to different regional growth in emissions. In the base case (IS92a), a uniform growth rate was used for the surface emission of pollutants. The sensitivity run was performed with the same global growth rate as in the base case but with the different regional growth rates given in Table 4-3.
Sensitivity to different projections of aircraft emissions. Instead of NASA 2015 emissions, the ANCAT-2015 aircraft emission data set was used. Total ANCAT emissions for the year 2015 are about 15% larger than the NASA emissions (see Section 9.3.4). Further differences relate to the location and seasonal variations of emissions.
Sensitivity to inclusion of NMHC chemistry. This study was conducted by making runs in which NMHC chemistry was excluded and comparing the results with those in which NMHC chemistry was included.
Sensitivity to neglecting heterogeneous chemistry on background sulfate aerosols. This sensitivity was estimated by comparing results with and without the heterogeneous nitrogen pentoxide (N2O5) + H2O reaction on aerosol. The aerosol surface area was derived from model calculations of the sulfur cycle. Hydrolysis of N2O5 on wet aerosol converts active NOx into the reservoir species HNO3, which is effectively removed by rain out. Hence, in the base case, less NOy is present in the free troposphere and LS, and emissions by aircraft are more effective in producing O3.
Sensitivity to interannual variability in meteorology.
Sensitivity to uncertainty in emissions in 2050. This study is carried out by using scenario G instead of scenario F from Table 4-4. Total NOx emission changes from 2 to 3.5 Tg N yr^-1.

The results of the sensitivity studies are presented in Table 4-6. The table presents the relative sensitivity r (%) of each process by comparing aircraft-induced increases in global O3 for the base case and the sensitivity case:

where O3,1 is the global O3 increase (kg) up to 16 km for the base case in 2015 (i.e., scenario D-C) and O3,2 is the same increase calculated for the sensitivity study (i.e., scenario D'-C'). The sensitivity studies show that increases in background NOx from lightning (sensitivity study 1) and different growth rates in surface emissions in different regions (sensitivity study 2) have only a slight impact on O3 perturbation from aircraft emissions. In both cases, O3 perturbations are reduced. This finding shows that O3 production in the UT and LS is limited by NOx, rather than by hydrocarbons. Interannual variability in meteorology (sensitivity study 6) and exclusion of heterogeneous removal of N2O5 in the models also led to a small change in global average O3 perturbation (sensitivity study 5). Excluding hydrocarbon chemistry (sensitivity study 4) would have a significant impact on O3, resulting in a smaller perturbation. Furthermore, there were significant differences in the results between the two models used to perform the study. Comparison of results with two different emission scenarios (sensitivity study 3) showed a noticeable impact on global O3 perturbation. Sensitivity study 7, which was set up to test the response of O3 perturbation to greatly increased NOx emission from aircraft (estimated upper limit in 2050), showed that the response is nearly linear and similar to what is computed for smaller NOx perturbations.

IPCC Homepage