Atmospheric Effects of Aviation: A Review of NASA's Subsonic Assessment Project (1999)

The accuracy of current calculations of aviation's impact on the atmosphere is restricted by insufficiencies in atmospheric data and inadequate representation of some key physical and chemical processes in models. The lack of a suitable measurement database currently prevents full testing of models by comparing their results with atmospheric observations. At present, no model can treat in detail all the important processes, which operate over orders of magnitude in spatial and temporal scales. Thus, a hierarchy of models, from box models to global-scale 3-D models, is being used to evaluate the impact of aircraft emissions on the atmosphere. Several types of models were used for the NASA assessment (Friedl, 1997) and the European assessment (Brasseur et al., 1997), as discussed below. Note that none of these models are presently able to effectively incorporate heterogeneous processes on aerosols or aerosol/cloud interactions.

Six chemical mechanisms were evaluated for the SASS interim assessment. All the models were initialized with identical upper-tropospheric chemical and meteorological conditions. Among the mechanisms considered, calculated rates of ozone formation from NO_x-catalyzed reactions in the upper troposphere agreed to within 2–15%. The differences among the models seemed to be related to discrepancies in photolysis rates, nitrogen speciation, and free-radical concentrations.

The ability of the global models used for the SASS assessment to represent rapid vertical transport has been evaluated by simulating the transport of radon,

and comparing the results to those obtained by non-SASS models. This comparison has shown that the quality of the assessment models is comparable to that of other chemical transport models, but also that major uncertainties still remain in current model representations of tropospheric processes, such as convective exchanges in the troposphere. An NO_y-tracer study has been used to evaluate large-scale transport in the upper troposphere. As noted in Freidl (1997), the results showed that a vertical resolution better than 1 km at the tropopause level is required to simulate properly the vertical distribution of emissions. When the results of the models used for the NASA assessment are compared with the limited number of available observations, it is apparent that those models all have difficulties in representing some of the features present in the observations.

The European assessment report discusses the results of the GIM (Global Interpretation and Modeling) project of the IGAC (International Global Atmospheric Chemistry) Program, in which the distribution of NO_x was calculated by ten different models. Large differences in NO_x distribution were obtained, reflecting the models' differing transport formulations and source strengths. Most of the models seemed to underestimate the abundance of NO_x and overestimate the HNO₃/NO_x ratio in most regions. This suggests an incomplete understanding of several mechanisms—for instance, heterogeneous chemistry on aerosol particles, or the role of peroxyacetyl nitrate (PAN) in the tropopause region. Our limited knowledge of the NO_x distribution in the free and upper troposphere makes it difficult to evaluate the accuracy of the models' simulated ozone production. Brasseur et al. (1997) also note that the models appear to represent rather accurately ozone distributions in remote tropical and mid-latitude regions, but they underestimate ozone concentrations in other regions, perhaps as a result of poor representation of troposphere/stratosphere exchanges.

The simulation results reported in both documents agree with prior studies in suggesting that aircraft emissions could result in an increase in concentrations of both nitrogen oxides and ozone in the troposphere. The aviation-related NO_x increase in the upper troposphere in the latitude band 30 to 60°N is estimated to be as high as 50%, with ozone increases of a few to 10 pptv (Brasseur et al., 1997). These current estimates of the potential impact of subsonic aircraft are smaller than the estimates of a few years ago, both because current estimates of the emissions are lower and because 3-D models tend to calculate somewhat smaller changes than 1-D or 2-D models.

The AEAP's Global Modeling Initiative (GMI) is expected to provide evaluations of the atmospheric impact of the subsonic fleet for future AEAP assessments. AEAP seems to have developed a relatively detailed plan for GMI development and to be making progress in implementing much of this plan. However, before the GMI participates in any assessment exercises, it needs to be carefully

evaluated; its results must be compared with those of other models and with available observational data. We note that the stratospheric-related modules of the GMI have participated in international model comparison studies; the tropospheric components of the GMI need to participate in such studies as well.

Intercomparisons involving the GMI should also include a full program of detailed sensitivity studies. Since the results of such studies tend to be highly model dependent, these exercises should involve a range of models beyond the GMI, and should be carefully coordinated. Sensitivity studies should be performed both with the GMI and with existing 3-D chemical-transport models to better evaluate the importance of certain processes that are still not well understood. Among them are:

• convective exchanges and cross-tropopause transport;
• particles in the troposphere and lower stratosphere and possible heterogeneous reactions; and
• partitioning within the different nitrogen species, and the importance of nitrogen reservoir species.

One issue of particular concern to the panel is that the modules being developed under the GMI do not yet incorporate aerosols and their effects, and that it appears unlikely that the GMI will be able to do so realistically in the next few years. If, during the lifetime of SASS, the GMI cannot be employed to address the heterogeneous processes related to aerosols, contrails, and clouds, the large commitment to the GMI on the part of SASS must be questioned. It may be that alternative, perhaps less complex, models that do include aerosols might be more appropriate. Given SASS's required assessment milestone in 2001, it appears urgent to identify the model type and the modeling approach that can best incorporate aerosol effects. A reallocation of some of the resources earmarked for the GMI should be considered, to permit such an effort to be undertaken immediately.

Very few global 3-D modeling studies of aviation's climatic effects have been attempted. Adding or removing ozone and/or water vapor in different vertical and horizontal regions in a GCM permit some assessment of the radiative impacts of these gas-phase perturbations. The inclusion of particles in the 3-D climate model is much more difficult, however. As noted earlier, the radiative effects of contrail particles may be very different from those of naturally occurring cirrus, because of the different mechanisms by which they are formed. Contributing to the difficulties of representing the effects of aerosols and contrails in 3-D climate models are the uncertainties in portraying natural clouds in GCMs. Likewise, the observation that aircraft emissions can ''activate" cirrus formation needs to be investigated more fully, and a methodology to include this effect in climate

models needs to be assessed. The present techniques used to represent clouds in GCMs are highly parameterized; this, in combination with the lack of a detailed treatment of the physics of convection, leads to highly model-dependent results.

Ponater et al. (1996) used a global GCM to examine the climate response to aviation-related water vapor and contrails. They found that the direct radiative perturbations due to water vapor were very small, but that contrails could have a significant effect on climate, with the strongest response occurring under mid-latitude summer conditions; however, a number of uncertainties involving cloud radiative interactions make the results somewhat preliminary. These uncertainties make it much more difficult to use standard 3-D climate models (such as those used to evaluate the climate response to increased atmospheric CO₂, CH₄, N₂O, and CFC concentrations) to assess the climate impacts of contrails.

It is not clear how the climate modeling efforts currently supported by SASS are linked to the other components of the SASS project. Specifically, the climate modeling studies being carried out at NASA-GISS need to be more closely coordinated with the GMI efforts. More generally, SASS needs to encourage its various modeling and experimental groups to compile and evaluate the hemispheric-scale physical and chemical perturbations from aviation with the specific aim of providing databases that are directly useful to climate modelers.