Interim Review of the Subsonic Assessment Project: Management, Science, and Goals (1997)

As a preliminary review of the SASS project's research topics and strategy, the NRC's Panel on Atmospheric Effects of Aviation has chosen to evaluate the scientific questions and program responses that appear in Table 1–2 on page 10 of NASA Reference Publication 1385, Atmospheric Effects of Aviation: First Report of the Subsonic Assessment Project (Thompson et al., 1996). Two of the "Aviation-Unique Topics" listed there will be the subject of a separate PAEAN report, An Interim Assessment of AEAP's Emissions Characterization and Near-Field Interactions Elements. The third, Operational Scenarios, PAEAN considers to be beyond its charge. In this report we confine our detailed evaluation to the "Atmospheric Science Topics". Nonetheless, we do have a few comments relative to the near-field interactions, which are included directly below. In each case the Table 1–2 entry is reproduced verbatim in italics before the discussion of that topic.

Question:

• Can fluid dynamics and/or chemical processes in aircraft wakes alter properties of engine exhaust products or their deposition altitude to significantly influence the background atmosphere?

Program Response:

• Develop efficient and accurate algorithms for thermodynamic, physical, and chemical properties of wake and exhaust products between the engine exhaust plume and the location where interaction is influenced only by background atmosphere.

• Couple models within situ and/or remote exhaust plume measurements using current aircraft platforms.

In general, the panel agrees with both the importance of this question and the programmatic responses as written. We add the caution, however, that any new study of a complex environment may yield unexpected results. For instance, no one anticipated the surprisingly high concentration of particles in the Concorde plume that was found in the October 1994 measurements (Fahey et al., 1995), so their size distribution was not measured and their radiative significance cannot be assessed. PAEAN also agrees that the number of new particles that form, evolve, and survive in the wake will be influenced by the ambient aerosol population, particularly the existing surface area. Both measurements (Clarke, 1993) and models (Shaw, 1989; Hegg et al., 1992) indicate that a higher pre-existing surface area suppresses new-particle nucleation, because of preferential condensation of precursors onto existing surfaces. Hence, measured particle concentrations in wakes (as observed by the Airborne Southern Hemisphere Ozone Experiment and Measurements for Assessing the Effects of Stratospheric Aircraft (ASHOE/ MAESA) programs, for instance) can depend on engine-combustion characteristics, emission products, the dynamic/thermodynamic environment, and the preexisting aerosol (which can vary in space and time). Aerosol nucleation and growth in aircraft wakes also appears to be sensitive to the concentration of ions emitted by the combustors into the nucleation zone (R. Turco, personal communication).

The implications of these effects may be subtle. For example, if a given mass of sulfuric acid produces large number concentrations of particles, they can be expected to be smaller and less effective as CCN or IN than if only a small concentration of larger particles were produced. It is thus important to characterize adequately aerosol particle size and composition as well as concentrations, both for true emitted species and for aerosols in the surrounding environment. This will be true not only for in situ measurements but also for interpretation and extrapolation of ground-level engine testing to operational high-altitude environments. Given the nonlinearities of many critical properties, careful characterization will be critical to the interpretation of observations and the modeling of effects.

Recently, some of the needed measurements were accomplished in SASS's SUCCESS Project (NASA, 1996). In discussions at a 1997 meeting, PAEAN members conveyed to AEAP their concerns about the diffuse nature of that project's measurement plans and the lack of adequate commitment to development of instrumentation critical to the project's goals. The panel would have more confidence that the proposals NASA solicited in NRA-96-OA-01 (3 June 1996) would indeed achieve the goals of the SUCCESS project if a tighter link were apparent between the key issues identified for near-field interactions, the

strategy proposed for addressing them, and the determination of the merit and appropriateness of proposed new instruments.

Question:

• What chemical and physical processes in the atmosphere could be perturbed by aircraft emissions?

Program Response:

• Use model sensitivity studies to identify chemical and radiative processes most likely to be perturbed or, in collaboration with models, place upper limits on minor processes.
• Identify chemical processes for gas, liquid, and solid phases that are affected by aircraft emissions.
• Determine rates of physical and chemical processes to guide observations and modeling.

The Atmospheric Effects of Stratospheric Aircraft (AESA) and Subsonic Assessment projects have supported several laboratory studies that have provided key kinetic information required for computer simulation of the effects of aircraft emissions on tropospheric and stratospheric ozone. The question quoted immediately above is indeed the crucial one, and the project responses will be important steps in the forthcoming SASS assessment.

For a number of years, NASA's Jet Propulsion Laboratory (JPL) and the International Union of Pure and Applied Chemistry (IUPAC) have made careful evaluations and reviews of kinetic data related to gas-phase reactions important in stratospheric chemistry. The latest of NASA's reviews appeared in 1994 (DeMore et al., 1994), and a new evaluation is being prepared at this time. These reviews provide valuable starting points for the suggested studies, but the greater complexity of the composition of the troposphere requires the evaluation of many more chemical reactions to account properly for the presence of additional non-methane hydrocarbons (NMHCs) and the reactive oxidation products that they form (volatile organic compounds, or VOCs).

The research plan suggested in Chapter 5 of the first SASS report (Thompson et al., 1996) is to extend the existing rate-coefficient evaluation to include the specific reactions of the C₃ hydrocarbons and their oxidation products (propane, propene, acetone, etc.) and to use a lumped-parameter approach for the higher-molecular-weight species (C₄ to C_4+n ). The need for additional evaluation of rate coefficients for hydrocarbons >> C₃ should be based upon a study of upper-tropospheric air analyses that are available in the literature (e.g., Blake et al., 1996), the range of concentrations of each species encountered, and the sensitivities of the ozone-generation steps to the concentration of each. The composition of the upper troposphere is far simpler than that of the polluted lower troposphere, and

evaluation and inclusion of the reactive hydrocarbons (RHs) >> C₃ may or may not be high-priority tasks.

The SASS report notes that heterogeneous chemistry and solution-phase chemistry are to be emphasized in the suggested research because they are less well understood than the gas-phase chemistry. This is reasonable in the long term, but achieving useful results within AEAP's time frame will be very challenging. Attention should probably be focused first on heterogeneous chemistry associated with types of particles expected to be of significance to cloud-nuclei formation and radiative effects, such as black carbon and sulfuric acid (see Stolarski and Wesoky, 1993). An activity in the area of aqueous chemical kinetics, as suggested in the SASS report, would also be of interest. Perturbations of ozone by aircraft emissions are most likely to arise from increases in tropospheric NO_x , SO₂ , SO₃ , and both H₂ SO₄ and carbon-rich aerosols. The extent to which H₂ SO₄ and other aerosols enhance contrail and cloud formation ultimately will determine the impact of aircraft emissions via heterogeneous reactions. If the results of current and future observations suggest that the amount of aqueous aerosol in the troposphere is significantly increased by aircraft emissions, then several potentially important reactions should be included in the evaluation studies, among them N₂ O₅ + H₂ SO₄ aerosol (Fried et al., 1994) and HO₂ + aqueous aerosols containing transition metal ions (Mozurkewich et al., 1987; Cooper and Abbatt, 1996). (See also the discussion of Johnston (1994) on the importance of including heterogeneous chemistry in modeling of the upper atmosphere.)

The reaction of N₂ O₅ on sulfuric-acid aerosols is now reasonably well characterized for a variety of surfaces, and its significant participation in ozone loss in the aerosol-enhanced stratosphere has been reasonably well established in modeling studies (Solomon et al., 1996). In addition, the nitrate/sulfate ion ratios in precipitation suggest that the reaction of N₂ O₅ to form HNO₃ is important in the troposphere (Calvert et al., 1985). The HO₂ radical removal at aerosol surfaces (Mozurkevich et al., 1987) has been confirmed recently to have an accommodation coefficient >> 0.2 on H₂ SO₄ surfaces (Cooper and Abbatt, 1996), and this could be an important loss process for HO₂ in aerosol-rich regions of the troposphere. To our knowledge, this reaction is not now included in any tropospheric models of the atmosphere.

The evaluation of the kinetics of potentially important aqueous-phase chemistry is suggested as part of the SASS research plan. The possible importance of solution-phase reactions within the troposphere has been given extensive consideration through the years (see, e.g., the discussion of Pruppacher et al., 1983; Hoffmann and Jacob, 1984; Schwartz, 1984; and Jaeschke, 1986). In November of 1993, AEAP held a workshop devoted to the discussion of heterogeneous and solution-phase chemistry that could be important to the SASS project. It is not clear which aspects of the recommendations made there have been implemented to date, but NASA's program for evaluation of these types of chemistry has continued to evolve. Recent modeling efforts have suggested a number of reac-

tions that may be of significance in the chemistry of the troposphere (see, e.g., Jacob, 1986; Lelieveld and Crutzen, 1990; Faust and Allen, 1992; Möller and Mauersberger, 1992; and Warneck, 1992). On the other hand, using different assumptions from Lelieveld and Crutzen about the water content of clouds, solubility of CH₃ O₂, and the HO₂ + HO₂ (H₂ O) >> H₂ O₂ (H₂ O) rate coefficient, Liang and Jacob (1997) have concluded that the effects of aqueous chemistry on summer tropospheric ozone in the tropics and middle latitudes are less than 3 percent. Together, this newer work and the significant wealth of older literature should provide the basis for deciding which reactions require further study. Among the reactions that should be considered is OH-radical generation through sunlight irradiation of aqueous aerosols containing iron salts in solution, which are ubiquitous throughout the troposphere (Graedel et al., 1985, 1986; Weschler et al., 1986).

At present, models considerably overpredict HNO₃ (by a factor of up to 5). It has been difficult to make reliable measurements of HNO₃ and NOy, and large uncertainties are associated with previous data. Recently developed techniques can provide rapid response and seemingly accurate measurements of HNO₃ in the upper troposphere (R.L. Mauldin III, personal communication), and more reliable HNO ₃ data will be forthcoming. NASA field programs should lead in the development and application of these improved methods. It is important to resolve the problem of this overprediction, since the failure of models to predict the observed NO_y components correctly will greatly affect the validity of assessments made with the GMI or other models.

An important element missing from the kinetic evaluation outlined in the SASS report is the program to reevaluate j-values (the apparent first-order decay coefficients related to photochemical processes) for the important light-absorbing species over the vertical extent of the troposphere, e.g., reactions of O₃ to form O(¹ D), of CH₂ O (formaldehyde) to form H, HCO (ultimately giving HO₂ ), CO, and H2 , and of CH₃ COCH₃ (acetone) to form CH₃ , CH₃ CO (ultimately forming CH₃ COO₂ and CH₃ O₂ ). The latest NASA evaluation of the j-value components (absorption cross-sections and quantum yields) (DeMore et al., 1994) does not include significant dissociation through light absorption of ozone within the long-wavelength tail, although current results of both theory and field studies (Michelsen et al., 1994; Shetter et al., 1996) suggest that this is significant. The ozone j-values recommended by DeMore et al, are thus probably incorrect. It must be remembered that the recent agreement seen between the j-values employed in the AEAP model intercomparisons (Stolarski et al., 1995) is proof only of the consistency of the model assumptions, not of the accuracy of the common choices of j-values now employed.

A major problem remains in lowering the uncertainties associated with j-value calculations in the presence of clouds and aerosols, conditions normally present in the real troposphere. Considerable progress has been made in delineating the expected effects in theoretical calculations (Lantz et al., 1996). Upward-

and downward-looking receivers that permit measurement of spectrally resolved solar radiation have been mounted on aircraft involved in NASA's tropospheric-observation programs. As expected, total irradiance varies greatly from clear-sky values when the aircraft fly above, within, and below clouds. The continuation of such measurements will be an important element in testing current theories and in arriving at reasonable algorithms to account for the effects of clouds and aerosols on j-values in tropospheric-chemistry models.

• Update not only the NASA and IUPAC evaluations of gas-phase rate coefficients, but those of j-value components for the photochemically active trace gases as well.
• Using existing representative chemical analyses of trace gases in the upper troposphere, together with measured (or extrapolated) rate coefficients, estimate the magnitude of ozone change that results when the chemistry of NMHCs > C₃ is included in the chemical model.
• If the estimated ozone change above is greater than a few percent, develop a lumped mechanism to simulate the chemistry of this group of "heavy" hydro-carbons.
• Estimate the average magnitude of the increases in NO_x , aerosols (surface area), and cloudwater (volume) expected as a result of aircraft emissions. If these estimated increases are relatively insignificant (less than a few percent), the models need to include only the currently recognized heterogeneous and solution-phase chemistry. If the expected effect is greater than a few percent, a significant effort should be expended to identify all possible heterogeneous reactions that could affect ozone levels in the troposphere. This study may involve largely modeling studies constrained by a combination of existing data and plausible uncertainty limits to bound the issues. The rate parameters for the reactions that appear to be potentially important in the modeling studies should be examined and tested carefully with well-planned, pertinent laboratory measurements.

Questions:

• What is NO_x conversion time and odd nitrogen partitioning in lower stratosphere and upper troposphere?
• What is aircraft contribution to upper troposphere NO_x budget compared to strat/trop exchange, lightning, and convective input of pollution?
• How reliably can NO_x and NO_y (gas and bulk phase) be measured in the upper troposphere and how does it affect the reliability of a NO_x budget assessment?

Program Response:

• Perform in situ measurements of tracers (e.g., CO₂ , N₂ O, CH₄)
• Plan NO_x and HO_x budget experiment(s).
• Support development work on NO_x and NO_y sensors.
• Assemble existing NO_x and NO_y database and study budget information from past UT/LS measurement campaigns.

This SASS table entry identifies the key questions correctly, if rather generally. The program response given in general terms is also adequate, except that the definition of the tracers to be measured in situ should be sharpened to read ''... of the tracers that help to identify the origin of upper tropospheric NO_x" For example, high O₃ but low CO and H₂ O would point to stratospheric air, high radon to continental surface air, and high H₂ O to air processed in moist convection and possible influence of lightning. It should be noted, however, that an attempt should be made to search for an unequivocal tracer for air processed through jet engines (e.g., soot). The panel does agree with SASS's philosophy of conducting process-oriented field campaigns, rather than continuous monitoring; the former promises immediately useful results within the existing. budget.

Assembly of a data base should include not only the species currently being followed, but also the molecules O₃ , H₂ O, CO, CH₃ COCH₃ , and other VOCs. It should also incorporate data for the upper troposphere from the space shuttle and satellites, as well as from MOZAIC, the European program for measuring O₃ and H₂ O from commercial aircraft. Some details of that research—but not enough for evaluation of progress—were given in the description of SONEX handed out at the January 1996 PAEAN meeting. In principle, aircraft campaigns with the duration, spatial coverage, and instrument mix outlined for the SONEX experiment could go far toward characterizing the relative contribution of NO _x from aircraft to the upper-tropospheric budget. But the main objectives of SONEX are not made clear in that writeup, and the sub-objectives are not prioritized. Some, but not all, of the sub-objectives can be achieved within one type of mission.

To allow proper evaluation of an experiment such as SONEX, and of progress toward SASS's goals, the main mission objectives need to be clearly stated. For example:

• Study upper-tropospheric NO_x and HO_x chemistry: Do observations confirm or deny present conceptions and model results? With what uncertainties?
• Study upper-tropospheric NO_x distribution and budget: Can contributions from different sources be quantified? What geographical locations and seasons have to be selected to demonstrate the efficiency of the various source processes, in particular lightning? What tracers correlate or anticorrelate with which sources?

A list of desirable sub-objectives should be provided, and the selected sub-objectives noted with a rationale for their selection. The priorities of each partial mission should be clearly stated, and the objective(s) it fulfills should be noted, so that all participants understand the reasons for the field director's decisions.

Other relevant questions are:

• What can the SONEX program learn from measurements made by SUCCESS and other AEAP-related experiments?
• How can the SONEX results be used for evaluation of the Global Modeling Initiative and other models?
• What information will SONEX provide on the lightning source?

Although SASS has a limited lifetime, in it NASA has an opportunity to leave a valuable legacy. Future research on changes in the chemistry and composition of the atmosphere would benefit greatly from a strategy for atmospheric monitoring to detect changes of significant magnitude or character in tropospheric chemistry. Public concerns over climate and environmental change can best be allayed by trustworthy data. SASS, with its involvement in both modeling and observations, is in a position to propose an appropriate strategy.

• Ensure that HO_x and all HO_x precursors and sinks currently thought to be of importance in the upper troposphere, are included in proposed measurement programs. Examples of additions needed are CH₃ COCH₃ (see Singh et al., 1995) and CH₃ OOH, methylhydroperoxide.
• Use model sensitivity analyses to identify the transport and chemical processes most important for the NO_x , HO_x and O₃ budgets in the upper troposphere. What can SONEX do to study these and reduce their current uncertainties?
• Seek model guidance as well for the placement, timing, and instrumentation of the missions. What kinds of model would be needed for that purpose? Are they available within SASS?
• Clarify the management structure, planning process, and operational implementation of SONEX. These steps should ensure better responsiveness to SASS's needs than is currently apparent.
• Consider looking beyond SONEX to participation in a coordinated program for long-term monitoring of tropospheric chemistry.

Questions:

• What are effects of contrails on Earth's radiation budget?
• Does aircraft exhaust affect ambient citrus properties?
• Do aircraft emit enough soot or sulfate to be radiatively significant?

Program Response:

• Satellites and ground based observations, supplemented by aircraft overflights.
• Focused aircraft expedition to sample ambient clouds, including chemical composition.
• Retrospective analyses of aerosol data, use of estimated emissions from aircraft, and future measurements.

The PAEAN panel agrees that climate forcing through direct and indirect radiative effects of aircraft effluents could be one of the most important potential consequences of aviation emissions. SASS's three questions listed above are relevant, but it would be well to reevaluate them in the light of the current concern with aerosols (NRC, 1996). The number, mass, and composition of emitted aerosols are expected to evolve over time as they go from the point of emission to a location at which they may influence the properties and microphysics of clouds. It will therefore be important to incorporate the evolution of aerosol emissions into assessments of possible cloud perturbations, both at aircraft operating altitudes and elsewhere in the troposphere.

The program responses shown above appear to be reasonable. Because PAEAN has heard most about the aircraft (SUCCESS) campaign, it seems appropriate to review it as an example of how SASS is executing its program responses. Chapter 7 in the first SASS report (Thompson et al., 1996) identifies radiative forcing as a "major area", and emphasizes that "experimental strategies will be developed so that process models are an integral part of the design, with each model selected to answer a key question defined in the mission plan." It notes further that this type of assessment will require prioritization of two types: the use of models and previous studies (including sensitivity studies) to focus measurement strategies on the highest-priority species, and determination of the conditions responsible for variability of selected trace gases, so that sampling aircraft will be flown in regions where the variability will be appropriately characterized. The chapter also notes that a 1994 workshop stated bluntly that "the sensor technology for making the most important measurements is either inadequate or nonexistent at this time.''

The panel supports this deliberate approach to identifying high-priority objectives and placing statistical bounds on the variability observed, as well as the recognition that the "radiative forcing" questions being addressed involve com-

plex issues in heterogeneous chemistry and subtle interactions with meteorological and dynamic environments, including water-vapor concentrations as well as detailed aerosol microphysical and optical properties. While direct injection of exhaust aerosol may at times have measurable radiative effects, the marked amplification of particle-light interactions when these particles act as CCN or IN contrails and/or clouds make this a particularly important and difficult area of study. PAEAN suggests that because of the relevance of its aerosol observations to possible climatic effects, SASS consider formulating objectives and missions that are consonant with the approach and recommendations outlined in A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change (NRC, 1996).

The inconsistency between the careful approach outlined in the SASS report and the approach of the SUCCESS campaign as described in its mission plan (NASA, 1996) is therefore surprising. The mission document provides an extensive listing of potential questions that might be addressed, but no effort is made to prioritize these questions or to establish key mission objectives. Indeed, the only prioritizing mentioned is identification of which SUCCESS mission type should be flown first. There is no mention of acceptable uncertainty limits for the measurements, or for that matter the mission objectives, let alone how they might relate to model requirements. Furthermore, there is no evidence that model results were used to either identify or "focus strategies on highest-priority species," nor does any assessment appear of the readiness, appropriateness, or adequacy of the selected instruments.

We caution that these considerations need to be reflected in the analysis or interpretation of SUCCESS data, and strongly recommend that future aircraft missions explicitly address these issues in establishing their science plans. For example, before even beginning to outline plans for exploring the aerosol topics mentioned above, the sequence below should be followed:

• Ask what needs to be known about the aerosols resulting from aircraft exhaust to (i) establish their direct radiative effects within a prescribed uncertainty, (ii) establish their indirect radiative effects through possible perturbations of CCN spectra, and (iii) establish their effects on actinic flux.
• Determine what kinds of coordinated measurements need to be made by ground-based, airborne, and spaceborne sensors to elucidate the questions posed above. On the modeling side, decide what kind of modeling should be done to determine the sensitivities of radiative effects to the changes in cloud optical properties caused by changing aerosol properties, and what kind of modeling is needed to help design effective field experiments.

Once these steps have been taken—and they should hold for all measurement campaigns, not just aerosols—the mission plan of any field experiment should state clearly how the results of the experiment will fulfill one of the needs identified.

• As part of the "retrospective analyses" mentioned above, use existing data sets (U.S. and other) to endeavor to establish background levels of aerosols and species of interest, and a history of the changes in both as far as they can be documented.
• As part of a more coordinated research strategy, require all principal investigators to demonstrate explicitly how, and how well, proposed work will meet the identified SASS needs and priorities, both in this area and for the project as a whole.
• As part of a more focused modeling strategy, determine what kind of physical parameters (e.g., refractive index, size distribution, shape) and theoretical developments are required to model the optical properties of the particles.
• As part of a more coordinated measurement strategy, determine what kind of measurements are required to establish the direct and indirect effects of aerosols from aircraft exhaust on radiative forcing and actinic flux.

Questions:

• What are predicted ozone changes and climatic impact associated with aviation?
• Can models explain observations?
• What are uncertainties in these predictions?

Program Response:

• Develop 3-D global chemical transport assessment model.
• Use global climate models and their embedded radiative models to evaluate the potential climate forcing from aircraft.
• Test models against atmospheric measurements.
• Model intercomparisons and error analysis, including subgrid processes and parameterizations.

The first question above is really a composite of two questions, and should be broken into (1) what are the predicted changes in ozone associated with present and future aviation? and (2) what is the impact of these changes and other aviation-related changes upon climate? They might be better stated as "What changes in concentrations of chemically and radiatively active constituents are predicted to result from various possible levels of aviation in the future?" and "What would be the climatic impact of these changes?" The constituents here would have to include all active species in or resulting from aircraft exhaust, from CO₂ to soot. The next closely related question would be how each of these constituents relates to potential changes in ozone, aerosols, cloudiness, and so on that affect climate. All these would need to be determined by the models as a

function of space and time, because the consequences of changes might be highly time- or region- specific. Modifications of stratospheric/tropospheric exchanges and the shifts in the tropopause level could also affect tropospheric (and stratospheric) climate. This first question is therefore extremely complex, even if SASS must currently restrict its investigation to determining the parameters needed to calculate the radiative forcing resulting from aircraft emissions.

The AEAP has begun to respond to this question by designing the Global Modeling Initiative (GMI), which has undertaken the development of a general-circulation-type chemical-transport model—composed of a core structure and various modules that simulate atmospheric processes such as transport and chemistry—that will reveal the effects of aircraft emissions on the chemistry of the troposphere and lower stratosphere. These modules, including input data, are submitted by contributing scientific groups. Such an approach looks promising, and the panel supports AEAP's decision to apply its resources to a single "community" model, rather than several separate ones. It is essential, however, that the GMI model and any others that are used as the basis for AEAP assessments be rigorously compared with available observations (concentrations as well as ratios of quantities). The meteorological data sets used to drive the transport of species, which come from either assimilation of observations (for example, those of the European Centre for Medium-Range Weather Forecasting or the NASA Data Assimilation Office) or from archives of general-circulation model data, should be carefully evaluated. Evaluation of results, with regard to both the transport of long-lived species and the distribution of chemical compounds, should be a high priority; in particular, the different advection, convection, diffusion, and chemical schemes used for the GMI should be intercompared in detail. For optimal comparisons, SASS needs to define which data sets at which time scales should be compared with model results.

Laboratory as well as modeling studies have shown during the past few years that heterogeneous-chemistry processes occurring at the surface of particles could be important for modeling the distribution of chemical species such as ozone. Although such processes are not well understood, their possible impact on global-scale species distributions should be evaluated. To this end, it would be desirable to establish an archive of observational aerosol data that would represent the troposphere. Important characteristics to include would be type, size, composition, and regional characteristics (e.g., regions of the troposphere where Asian or Saharan dust is commonly found, or products of biomass burning). Such a data base could be used in conjunction with known aircraft emission characteristics to guide laboratory studies of heterogeneous chemistry on aerosols, or provide realistic aerosol scenarios for modeling studies.

Currently existing three-dimensional (3-D) chemistry-transport models (including those funded through other programs) should be used to identify and evaluate key issues such as NO_x budget, tropospheric heterogeneous chemical processes, deep convection, and stratospheric/tropospheric exchange. Perform-

ing sensitivity studies with models that have lower temporal and spatial resolution will help evaluate the main processes that determine the impact of aircraft, and define the uncertainties associated with these processes. Such evaluations should be regarded as essential, and should also be used in designing new observation campaigns and assessment strategies that will help reduce the uncertainties.

The GMI effort is mostly concerned with evaluation of impacts of aircraft emissions on the distribution of gaseous chemical species (and ultimately aerosols) and calculation of radiative forcing resulting from these changes. As was discussed in the first SASS report (Thompson et al., 1996), however, the GMI is not intended to evaluate possible climate perturbations resulting from changes in forcing caused by the effects of aircraft. Possible feedbacks between temperature and chemical-species distributions (which are nonlinear) will not be taken into account, and perturbations of the climate system resulting from indirect impact of particle emission or formation will not be evaluated on the global scale. While changes in temperature could further alter O₃ concentration, the resulting effects are expected to be small, so neglect of the second-order effects can be justified at this stage. Considerably more resources would be required for SASS to evaluate climate impact; SASS researchers can instead take advantage of ongoing related efforts in the United States and abroad. For example, extensive work in the area is performed under the aegis of the World Climate Research Programme (WCRP).

The Global Modeling Initiative is unique to NASA, and has the potential to be a valuable vehicle for modeling other global environmental processes in the future. Because the GMI model will not be fully coupled to any general-circulation model for the next few years, the impact of changes in the radiative balance caused by aircraft emission of either gas-phase species or particles will have to be evaluated non-interactively by using existing climate models. Perturbations of ozone concentration or of particle distribution calculated by the GMI model could be used as the input to such climate models. These first evaluations could be performed through collaboration between AEAP or SASS research groups and climate modeling groups at institutions such as NOAA's Geophysical Fluid Dynamics Laboratory, NASA's Goddard Institute for Space Studies, and NSF's National Center for Atmospheric Research.

The GMI model is expected to be an important contributor to the next SASS assessment. The development of this model represents a large amount of work, and it implies collaboration of many scientific groups. For successful completion of this initiative, it is important to define a detailed plan for the GMI components' development and integration, as well as for the model's evaluation and integration, and to estimate the computational and support resources required and available. The timeliness of the modeling results is a concern, and when funding cuts are a possibility, some strategic planning may be needed to ensure that efforts are applied in the most cost-effective areas.

• First, define a detailed time scale for GMI model development, testing and evaluation (including the aircraft-specific components), use for assessments, and coupling with other models.
• Put a high priority on GMI model evaluation, through comparisons of its results with those of separate modules, with available observations, and with results of other existing 3-D models.
• Perform chemistry and transport sensitivity tests with existing 3-D chemical-transport models. Where forcing terms appear to be small, as that for ozone does at present, perform sensitivity computations to test the model results.
• Use emissions scenarios, recent measurements, and the results of 2-D models that treat aerosols to assess what ranges of aerosol properties are important to models (both 2-D and 3-D), and set boundaries that can be incorporated into the GMI.
• Use the results of model sensitivity studies in designing new experiments—for example, the necessary spatial frequency of sampling—and define what is needed for the results to be most useful to the model.
• Develop stronger interactions with the U.S. and international climate communities to obtain estimates of the effects of the projected emissions on climate; determine how computations of changes in greenhouse gases and in aerosols could be used as input for climate models.