The Atmospheric Effects of Stratospheric Aircraft Project: An Interim Review of Science and Progress (1998)

Before an informed assessment of the potential impacts of HSCT emissions on the atmosphere can be made, a number of scientific issues need to be resolved, or at least better bounded. The manner in which emissions are transported horizontally and vertically away from the flight path needs to be better understood and modeled. The formation of aerosols in the aircraft's wake, and the heterogeneous chemistry that takes place on those new aerosols, require further exploration; reactions in the near field and in the polar regions are of particular concern. Atmospheric observations, together with development and refinement of both wake and global models, are called for to satisfy these goals. This chapter reviews the issues that, in PAEAN's judgment, deserve the highest priority within AESA.

Correct representation of global-scale horizontal and vertical transport processes in the stratosphere, and of stratosphere-troposphere exchange, is essential for assessing the effects of HSCT aircraft emissions. Two-dimensional chemical-transport models will continue to be the primary tool for evaluating the impact of stratospheric aircraft emissions at least through this year, when a preliminary assessment of potential HSCT effects is expected.

Unfortunately, these models cannot yet treat the transport very realistically. As explained in Chapter 2, after global atmospheric forces have pushed air up into the tropical stratosphere, much of that air fans out in the mid-latitude middle stratosphere (where the ozone distribution is dominated by ultraviolet radiation

and chemistry), and some air is transported by horizontal motions into the polar winter night (where the ozone distribution is determined by dynamics, and by the emission and absorption of infrared radiation) and then down to lower altitudes. Some of this descending air spreads out over the lowest ''overworld" to constitute the region of maximum ozone density; some of it continues to descend into the lowermost stratosphere and then exits into the troposphere.

If a fleet of Mach 2.4 aircraft were to cruise at about 20 km altitude outside the tropics, approximately two-thirds of the emissions would be transported down into the region where ER-2 measurements and the current understanding of heterogeneous chemistry suggest that HSCT NO_x would have only minor effects on ozone. Only one-third of the emissions would remain in the region in which NO_x is active in reducing ozone. Of the emissions occurring at tropical latitudes (estimated to be more than one-third of total emissions in 2015), however, much will be transported by the tropical updraft into the region where NO_x dominates ozone destruction.

The AESA assessment models need to incorporate these atmospheric transport processes realistically if their results are to be useful for evaluating the potential impact of an HSCT fleet. Improvement of the models' transport parameterizations is needed, which in turn will require a database of the radicals and species at the relevant altitudes. Further measurements, and continuing analysis of existing data, will be necessary to reduce the existing uncertainties, particularly above 20 km and in tropical regions.

Since late 1978, instruments aboard circumpolar NASA satellites have measured global ozone and its vertical distribution every day over the sunlit portion of Earth. Since the late 1950s, between 50 and 100 ground-based Dobson meters have measured local total ozone at unevenly spaced locations, mostly in the northern hemisphere. Figure 9 presents zonally averaged changes, measured and modeled, in total ozone between 1980 and 1990 as a function of latitude. The group of models, as a whole, show the same general arching pattern as the observations from pole to pole, and agree well in the low and mid-latitudes with each other and the observations. At the north pole the model-to-model comparisons vary by a factor of 3 (about 2 to 6.5 percent loss per decade), however, and at the south pole they vary by a factor of 5 (about 2 to 9.5 percent). These differences indicate that the assessment models do not yet represent polar processing realistically, and may lack or misrepresent other features of the chemistry or dynamics.

As noted in Chapter 2, the behavior of ozone in the polar regions may be worthy of special attention by AESA. Large seasonal decreases in ozone have been observed over Antarctica and the Arctic region, and it is known that heterogeneous processes play a key role. But ability to model particle-formation and

FIGURE 9

Changes in annual average ozone column from 1980 to 1990, as calculated by  nine models, together with the trend derived from three sets of observations.  MPIC is the model of the Max-Planck Institute for Chemistry; TOMS and SBUV are  data from the Total Ozone Mapping Spectrometer and Solar Backscatter UV  Spectrometer respectively. (From WMO, 1995; reprinted with permission of the  World Meteorological Organization.)

-growth processes quantitatively remains limited, and the related uncertainties are considerable. This constraint is significant in the HSCT context, because attempts to incorporate polar processes into assessment models have shown that the ozone response to HSCT emissions is sensitive to the representation of particle microphysics (see Pitari et al., 1993, and Stolarski et al., 1995). Recent studies by De Rudder et al. (1996) and Tie et al. (1996), which include more realistic descriptions of particle microphysics, indicate additional depletion of ozone by HSCT emissions when polar stratospheric clouds are included. Also, recent modeling of PSC formation (Del Negro et al., 1997) indicates that additions of NO_x, aerosol, and water vapor in amounts characteristic of the assumed 500-craft HSCT fleet can enhance polar aerosol volume on the order of 100 percent for temperatures below 192K. This effect raises the possibility of substantial additional ozone depletion in the polar regions in response to HSCTs via heterogeneous processes involving PSC particles.

Furthermore, ozone in the polar regions appears to be sensitive to interannual fluctuations of temperature and aerosol to a degree that may indicate particular sensitivity to HSCT emissions. Empirical evidence for this sensitivity comes from interannual changes in ozone depletion caused by halogen compounds. Both increased aerosol concentrations (due to a volcanic eruption) and lower winter stratospheric temperature have apparently enhanced the depletion caused

by halocarbons, which suggests that particular attention should be paid to aerosol resulting from HSCT emissions in the polar context (see Portmann et al., 1996). HSCT-related changes in NO_y or aerosol concentrations can affect the formation of PSCs, which in turn can alter the mechanisms and the temperature sensitivity of ozone-depleting processes.

HSCT exhaust could affect the polar regions not only as a result of transport from lower latitudes, but directly. Some projected routes, such as those between Europe and Alaska, pass directly through regions where the Arctic vortex is frequently established. In the Arctic region, reactions on aerosols and associated denitrification of the local stratosphere can cause rapid destruction of ozone during winter. An HSCT fleet could add NO_x, water vapor, and sulfate aerosol into the polar stratosphere, causing either reduction or enhancement of seasonal ozone loss already occurring due to halocarbons. Interactions among the NO_x, HO_x, and ClO_x chemical cycles in the gas phase mean that the NO_x in the exhaust may decrease rather than increase polar ozone depletion under some circumstances (Considine et al., 1994, 1995).

While HSCT emissions at non-polar latitudes may influence polar chemistry, as noted earlier, polar processes may also alter the effect of HSCT emissions occurring at non-polar latitudes. Polar processes shorten the atmospheric residence time of NO_x emitted by HSCTs (Pitari et al., 1993; Considine et al., 1994), which generally leads to lower calculated ozone depletion from HSCTs at northern mid-latitudes. Clearly, both further measurements and further modeling studies are needed to improve our understanding of PSC formation and evolution, and how those processes could be influenced by HSCT exhaust.

Modeling is a crucial aspect of AESA's effort to determine the atmospheric impact of a fleet of HSCTS. It provides the means of turning what is known about the atmosphere into predictions of what could become of it. Atmospheric measurements supply boundary values and initial conditions for models, laboratory measurements supply numerical coefficients for established physical and chemical equations, and atmospheric dynamics supplies concepts and equations for general 3-D motions in the atmosphere, which modelers approximate and parameterize in their 2-D and 1-D models. AESA's Models and Measurements study, described in the 1992 three-volume M&M report (Prather and Remsberg, 1993), compared the predictions of 14 different 2-D or 3-D models against a large body of selected atmospheric measurements. The spread among the models for most tests in M&M was comparable to that shown between 30°S and 60°N in Figure 9; no single model agreed with the measurements at all times.

AESA has a second Models and Measurements study under way for the 1997–1998 period. Of more concern to the PAEAN panel, however, are AESA's current efforts in the area of 3-D modeling. The NRC AESA Panel's earlier

evaluation (NRC, 1994) recommended that increased 3-D modeling have a high priority, particularly given 3-D models' value for sensitivity studies. Not long thereafter, the AEAP initiated the Global Modeling Initiative (GMI), sometimes referred to as the "core 3-D model" effort. There was insufficient time for this effort to have yielded any results by the time of NASA's 1995 assessment, but it is clear that 3-D representations of transport effects are very important for understanding the chemical composition and climate of the lower stratosphere, and for evaluating the impact of the proposed fleet of HSCTs. The use of the GMI will be valuable to future assessments, particularly once it can include aerosol-related chemistry and processes.

The sensitivity of the ozone balance to horizontal transport in the region of transition between chemical and dynamical control places strong constraints on assessment models. It is not clear that parameterized transport in 2-D models can realistically compute the evolution of the ozone mixing ratio in regions where the balance between chemistry and dynamics will be highly dependent on whether transport takes place by rapid, large-scale wave breaking (in which air parcels move through large latitudinal ranges in a few days) or by slow meridional diffusion (which has a seasonal time scale); many models' transport parameterizations reproduce only the latter. Whether such transport details are significant for the overall ozone balance in the assessment models can probably be answered only through careful studies with 3-D chemical-transport models.

The GMI is based on the "science team" approach that has served well in research satellite missions and in many field campaigns. It is not at all clear, though, that this approach will be as effective in the modeling area. Successful field campaigns (e.g., AASE, ASHOE, and SPADE) have featured not only strong leadership and dedicated participants but also sufficient resources to accomplish their goals. The GMI science team, however, consists of many groups or individuals for whom the GMI is a relatively minor commitment. Although the goals of the GMI science team for the AESA assessment have now been formulated fairly specifically, the panel is concerned that the model may not be able to provide 3-D integrations that will fulfill the requirements of the 1998 stratospheric assessment. A fully realistic 3-D model is further off still. The value of continuing the GMI effort, however, even if its results benefit only later assessments, is not in question.

Enhancement of AESA's plume/wake modeling effort was one of the earlier NRC panel's recommendations. Better understanding of the microphysics in the wake region is needed before the larger-scale models can properly represent HSCT effects. Two aspects of exhaust behavior and dispersion are of particular interest.

The first is important for NO_x reduction of ozone. The Mach 2.4 HSCT will

cruise between 18 and 21 km altitude, and inert material deposited at that altitude has a stratospheric residence time of about two years (Kinnison et al., 1994a). In the sunlit stratosphere, NO and NO₂ reach a photochemical steady state within a few minutes, and NO_x attains photochemical steady state with HNO₃ within a few weeks. After the exhaust gases spread, the mixing ratio of NO_y from the exhaust is a thousandfold or more lower than that of ozone. Only cyclic processes such as reaction (1), occurring hundreds of times, change ozone significantly, so the details of the reactions of NO _y in the wake of the aircraft appear to be of little consequence with respect to ozone. However, if some form of NO_y is permanently incorporated into an atmospheric aerosol, which should have a stratospheric residence time of about 1 year (compare Kinnison et al., 1994a), then the detailed behavior of exhaust components in the wake becomes more important for ozone.

The second aspect of exhaust behavior that is of concern is the formation of sulfuric-acid particles in the wake. Fahey et al. (1995) measured aerosol concentrations in the exhaust of a Concorde SST in flight using instrumentation on board the ER-2. From the fuel's known sulfur content and the results of particle-counter measurements, it was inferred that a large fraction of the sulfur in the fuel was converted to SO₃ before leaving the engine, and new particles formed in the near wake before being intercepted by the ER-2. The measurements indicated that 12 to 45 percent of the total fuel sulfur was rapidly oxidized to sulfur trioxide, instead of appearing as sulfur dioxide. Uncertainties related to the instruments measuring particle sizes, however, made it necessary to use crude size estimates in converting the number of particles measured into inferred particle mass (Fahey et al., 1995). These percentages thus should be viewed with caution, and evaluations of potential HSCT impact need to include the related uncertainties about surface area available for heterogeneous reactions and other details of the relevant microphysical processes.

The high aerosol concentrations observed in the wake may have been unexpected, but their formation from a high-temperature combustion process should not be considered surprising. Sufficiently high supersaturation in the presence of suitably low available surface area for condensation is likely to result in rapid nucleation of particles. Similarly high concentrations of particles have been observed for natural, low-temperature processes in the troposphere and stratosphere over far more extensive regions. Clarke (1993) reported that concentrations equal to those measured in the Concorde's wake occur over regions of several hundreds of kilometers at altitudes of 8–10 km. Later observations near the tropical tropopause showed similarly high concentrations of small (circa 15 nm) nuclei; it has been argued that they provide a source of nuclei upon which oxidized sulfur gases in the stratosphere can condense (Brock et al., 1995). Hence, interpretations of the significance of high concentrations of aircraft-derived nuclei need to recognize that extensive natural sources of similar particles can be present at the altitudes at which aircraft operate.

Many reactions between atoms, radicals, and molecules are at chemical equilibrium under combustion conditions. High-temperature equilibrium product distributions in the engine can be ''frozen out" during the rapid expansion and cooling of the combustion gases, and depending on the cooling rate and complex chemical kinetics of the system, the exhaust gases could contain significant amounts of gaseous sulfuric acid as well as gaseous SO₃ and SO₂. It appears that much of the sulfur in the exhaust forms fresh aerosol in the near wake (Kärcher and Fahey, 1997; Taleb et al., 1997), rather than accreting on existing atmospheric particles, which is consistent with much of the sulfur in the exhaust being in the form of SO₃ or H₂SO₄. This formation of aerosol would make a larger-than-expected surface area available for heterogeneous reactions. Weisenstein et al. (1996) used a 2-D model to calculate the change in annual average ozone column at 57°N caused by the emissions of a Mach 2.4 HSCT fleet. They assumed that the fleet would have an EI(SO₂) of 0.4. When they varied the percentage of sulfur emitted as particles (or particle precursors), the available aerosol surface area changed as shown below:

Sulfate aerosols at the UNEP level are known to have a strong impact on stratospheric chemistry by providing a site for the hydrolysis of N₂O₅. They may also be found to have a significant global impact on radiative forcing or other atmospheric properties. The consequences of an aviation-related increase in sulfate aerosols, in the context of these existing processes, must be considered. The possibility that aerosols from the exhaust might be transported into the troposphere, where they could affect cloud properties, should also be explored. Detailed modeling of aerosol particle formation and growth in the wake will be an important tool for determining the critical properties that govern the formation and the effects of the sulfate aerosols in the plume.

AESA has taken an initial step toward evaluating the impact of 500 supersonic aircraft flying at Mach 2.4 with an EI(NO_x) of 15 on climate, using the

Goddard Institute for Space Studies (GISS) general-circulation model (Rind and Lonergan, 1995). The radiative effects of changes in the vertical distribution of ozone and water vapor were evaluated separately. The ozone perturbations were found to cause a stratospheric cooling of a few tenths of a degree C, and a global mean surface-temperature increase of 0.025°C. This surface-temperature perturbation is not statistically significant when compared with the 0.09°C interannual standard deviation of surface temperature in the model, although a predicted high-latitude (70–90°N) stratospheric cooling of close to 0.5°C does appear to be statistically significant by comparison with model variability. The increase in water vapor in the stratosphere for the same 500-aircraft scenario has a similar effect: a stratospheric cooling of 0.5°C. The resulting mean surface-temperature increase (0.01–0.02°C) again is small compared with the model's standard deviation.

Soot and sulfate aerosols could also affect climate by altering radiative forcing in the atmosphere. Stratospheric sulfate aerosols may also contain substantial amounts of non-volatile solid particles (Farlow et al., 1977, 1978); although these solid components could affect the aerosols' radiative forcing properties, little work has been done to characterize their chemical composition. Other properties, such as size distribution, lifetime, and chemical reactivity, affect aerosols' climatic impacts as well. Certainly much remains to be understood before the potential impacts of increases in aerosol concentrations can be well quantified.

The NRC's AESA panel noted that NASA's 1993 HSRP Interim Assessment Report concentrated on the NO_x-ozone issue, and that no assessment of climatic impacts was included (NRC, 1994). They suggested that AESA should "conduct an immediate first-cut assessment" of possible effects of HSCT effluents on climate, including the modification of the thermal balance of the atmosphere by changes in the vertical ozone distribution, the alteration of the number or size distribution of cloud-or ice-nucleating aerosols, and the influence of added water vapor on cirrus clouds and contrails. From the research summarized above, it does not seem that significant impacts on climate from HSCT exhaust can be expected, although many uncertainties remain. The influence of contrails on cloudiness, a growing concern in the troposphere, is a very minor concern for AESA; over most of the stratosphere, HSCT aircraft emissions will not form contrails (Stolarski et al., 1995). (In the winter polar stratosphere, however, the HSCT exhaust constituents can contribute to the formation of PSCs, as noted earlier in this chapter.) Given the extreme complexity of accurately modeling the climate system, this panel does not recommend that AESA undertake further studies in this area.

As Stolarski et al. (1995) note, the uncertainty related to gas-phase reaction-rate coefficients is the only readily quantifiable uncertainty associated with the

various models used. The differences among the model results shown in Figures 8 and 9 suggest that the overall model uncertainties are considerably larger than those in the gas-phase rate coefficients alone. The spread of predicted ozone-column reductions among the nine models in Figure 9 sets a lower limit for the uncertainty of model-predicted ozone-column reduction. For example, at tropical latitudes, the observed ozone column decrease is 1±1 percent, and the models agreed with observations and among themselves within this range. At 60°S, however, the observed column decrease is 9±2 percent, and the models calculate a decrease of 3±1 percent. At the poles, the model uncertainties are even greater.

Similarly, in Figure 7 five AESA assessment models exhibit good agreement for calculated change of ozone vertical column, with a maximum spread of 1 percent. Figure 8, however, shows the uncertainty in the vertical distribution of the calculated changes in ozone concentration at 45°N. The models' vertical profiles agree fairly well among themselves at altitudes above 30 km, but they are noticeably different below that. For example, two models predict noticeable ozone production in the troposphere as a result of HSCT operation in the stratosphere, and two models give almost zero ozone production. These differences would yield different calculated warmings of Earth's surface. An ozone increase in the troposphere from HSCT exhaust would also affect ozone-change calculations for subsonic aircraft in the troposphere.

Chemical details not predicted by the models may also be quite important. For example, what if reaction (6) had turned out to be slow on sulfuric acid aerosols in the mid-latitudes, and reaction (8) had been fast? As shown by net reactions (12), reaction (6) leads to removal of active NO_x, but it does not directly increase the number of active chlorine species. However, the reaction sequence below (which includes reaction (8)) yields two units of active ClO, and converts one unit of active NO_x into nitric acid:

In the lower overworld and in the lowermost stratosphere, one unit of ClO is more active in destroying ozone than one unit of NO₂. In this hypothetical case, the addition of heterogeneous reactions to atmospheric models would have resulted in higher calculated ozone reductions from HSCT emissions in the mid-latitudes than those actually calculated.

This discussion illustrates the value of AESA's efforts to obtain a more complete understanding of the important features of stratospheric science. However, many uncertainties—for instance, those attached to transport processes,

microphysics, and heterogeneous chemistry—will be challenging to resolve. AESA needs to identify the most critical uncertainties, and develop a plan establishing research priorities that will reduce those uncertainties as much as possible during AESA's lifetime. Such a plan will permit response and input to these activities from a wider scientific community, and may enable the international atmospheric-science community to cooperate on critical research issues.