Aviation and the Global Atmosphere

	Other reports in this collection

4.3. Model Studies of Supersonic Aircraft

This section contains an evaluation of the effects of supersonic engine effluent on stratospheric O3, NOy, H2O, and other trace constituents. The assessment of the potential future supersonic aircraft fleet was conducted primarily with 2-D and 3-D CTMs. These global models are designed to represent the dynamic, chemical, and physical processes of the stratosphere. Chapter 2 discusses the validity of these models to accurately represent the present atmosphere.

Because these future aircraft have not yet been designed, we treat our study of supersonic aircraft in a parametric way. This assessment performed calculations for more than 50 supersonic scenarios and examined variables such as EI(NOx) and fleet size, ambient SAD and chlorine radical abundances, SO2 gas-to-particle conversion in the aircraft plume, cruise altitude sensitivity, and model cold aerosol chemistry representation. Results from this section are incorporated in Chapters 5 and 6.

4.3.1. Models Used in Supersonic Aircraft Assessment

This section presents calculated responses of O3 to HSCT aircraft from nine numerical models of the stratosphere. These nine modeling groups have been responsible for most recent publications on the subject. We restrict our analysis to this set of results because it is difficult to compare model results unless the models are produced in the same way. For example, it is difficult to compare the results here with recent publications of Dameris et al. (1998) because they isolated the effect of NOx emission by excluding other emissions from their simulation.

The six 2-D models and three 3-D models are listed in Table 4-7. The 3-D models obtain 3-D distributions of the species by solving 3-D mass continuity and chemistry equations. The 2-D models calculate the zonal-mean concentrations of the species by solving zonal-mean mass continuity equations. Transport of the zonal-mean concentration is affected by zonal-mean (vertical and horizontal) winds and eddy diffusion fluxes that simulate the effects of zonally asymmetric motions. Chemical packages are used to compute zonal-mean production and loss rates for each species. In this section, we highlight some of the differences in model formulations that may have contributed to differences in model predictions. The reader is referred to the Technical Report on Supersonic Aircraft Effects for additional details.2

4.3.1.1. Domain and Resolution

The lower boundary of the SLIMCAT model is at 335 K potential temperature (~ 10 km). All other models have lower boundary at the ground. The top boundary varies from 60-90 km. Vertical resolutions in the lower stratosphere range are 1.2 (AER), 1.5 (LLNL, SCTM1), 2 (GSFC, CSIRO), and 3 km (UNIVAQ, LARC, THINAIR, and SLIMCAT). Horizontal resolutions for the 2-D models are 5� latitude for LLNL and CSIRO and 10� latitude for the rest. For the 3-D models, horizontal resolutions are 5.5�x5.5� for LARC, 7.5�x7.5� for SLIMCAT, and 10� longitude x 7.8� latitude for SCTM1.

All of the models use temperature lapse rates to define the location of the tropopause, which changes with season and latitude. Most 2-D models use temperature from climatology, whereas the 3-D models use temperature from GCM output or objectively analyzed temperatures that are based on measurements-for example from National Centers for Environmental Prediction (NCEP) or European Centre for Medium-range Weather Forecasts (ECMWF). The troposphere is distinguished from the stratosphere in the 2-D models by assigning large values of horizontal diffusion coefficient (Kyy typically 1-1.5x106 m²/sec) and vertical diffusion coefficient (Kzz ~4-10 m²/sec) to simulate rapid mixing. Studies indicate that stratosphere-troposphere exchange may be dominated by transport from the mid- and high-latitude lower stratosphere to the troposphere (Eluszkiewicz, 1996). In a 2-D model, this transport will manifest itself as eddy flux along isentropic surfaces across the tropopause boundary (Shia et al., 1993). If this premise is true, calculated residence time of aircraft emissions in the LS would be sensitive to the horizontal and vertical resolutions of the models because resolution constrains the location and seasonal variation of the tropopause.

4.3.1.2. Temperature, Transport Parameters, and Solvers

The THINAIR model is the only model that calculates temperature and transport circulation consistent with calculated O3. The rest of the models generated results in CTM mode; in other words, results are obtained using pre-calculated temperature and transport fields. Different methods are used to compute residual mean circulation and eddy diffusion coefficients. Transport circulation for the UNIVAQ 2-D model is taken from a low-resolution spectral GCM (Pitari et al., 1992). Winds and temperature for the LARC and SCTM1 models are from off-line GCM simulations, and those for SLIMCAT are from the UKMO analysis. There is no accepted method to validate computed transport parameters. Temperature is used to compute temperature-dependent reaction rate constants and, in some models, to predict the surface areas of PSCs. Different numerical schemes are used to solve the mass-continuity equations. Given the different methods used in deriving the transport parameters in the models, it is not surprising that there are large differences in calculated distributions of trace gases in the models. Large differences in model-simulated distributions of chemically inert tracers such as sulfur hexafluoride point to transport differences as a major contributor. The research community is trying to identify a climatological database for zonal-mean distributions of trace gases through an ongoing exercise (Models and Measurements II, Park et al., 1999) that can be used to diagnose transport parameters.

4.3.1.3. Chemistry

Previous model intercomparison exercises have shown that chemistry solvers in most models calculate the same partitioning of radicals under the same constraints (solar zenith angle, overhead O3, local temperature, local sulfate surface area, and local concentrations of the long-lived species) when they are used as box models in photochemical steady-state. Because the 2-D models transport zonal-mean concentrations, zonal-mean production and loss rates are needed in the mass-continuity equations for long-lived species. Different techniques are used to obtain zonal mean production and loss rates, including integrating diurnally varying concentrations of radicals obtained by explicit time marching to compute the zonal-mean rates or using diurnally averaged radical concentrations calculated from average solar zenith angles corrected by pre-calculated correction factors.

Reaction rate constants are from DeMore et al. (1997). There is no recommendation in DeMore et al. (1997) for the yield of hydrogen chloride (HCl) from the OH+ClO reaction. Lipson et al. (1997) recently measured a temperature-dependent yield of 5-6%. Most models assume that the HCl yield from the OH+ClO reaction to be 0%. However, the CSIRO, LARC, and SLIMCAT models assume a 5% yield of HCl from OH+ClO. Because of the nonlinear dependence of reaction rate constants on temperature, the zonal average reaction rate constant calculated using local temperature as the air parcel moves around the globe zonally is different from the rate constant calculated using zonal-mean temperature. Some 2-D models chose to account for this effect by using a zonal temperature distribution based on observations; other models just used zonal mean temperature.

The following heterogeneous reactions were identified in DeMore et al. (1997) as possible reactions on sulfate or PSC. Most models assume that the rate constant is in the form gvA/4, where g is the reaction probability, v is the thermal velocity of the reactant, and A is the surface area of the sulfate or PSC.

N2O5 + H2O 2HNO3

ClONO2 + H2O HNO3 + HOCl

BrONO2 + H2O HNO3 + HOBr

ClONO2 + HCl HNO3 + Cl2

HOCl + HCl Cl2 + H2O

HOBr + HCl BrCl + H2O

These reactions occur on the surface of particles (sulfate or PSC). In addition, all but two of the nine models (LLNL and UNIVAQ) include the reaction N2O5 + HCl ' HNO3 + ClNO2 on PSC. In most cases, H2O and HCl molecules are adsorbed or dissolved in the particles, and the reaction proceeds as the other reactant collides with the particle. There are slight variations in the temperature dependence of g for reactions on sulfate particles used in the various models. Previous analyses (Murphy and Ravishankara, 1994; Borrmann et al., 1997; Michelsen et al., 1999) have shown that g is very sensitive to temperature for some reactions. As the air parcel goes around the globe, it experiences temperatures that are both lower and higher than the zonal mean temperature. Ignoring zonal asymmetry in temperature by using zonal mean values in calculations may underestimate the O3 impact from HSCT (Pitari et al., 1993; Considine et al., 1994; Groo� et al. 1994; Weisenstein et al., 1998). Because the sulfate surface area is specified in the calculations, the temperature-dependence comes from g and v. Some models compute an effective gv based on a temperature-weighted g and v from zonal-mean temperature, whereas other models compute a temperature-weighted product of gv. Some of the listed reactions also occur on type I PSCs (assumed to be nitric acid trihydrate (NAT)) or type II PSCs (assumed to be "ice"). Exact treatments in each model vary; see the Technical Report on Supersonic Aircraft Effects for details.

4.3.2. Definition of Scenarios

This section describes the scenarios for supersonic aircraft in 2015 and 2050 that were evaluated for this assessment. The premise for the future supersonic (proposed hypothetical HSCT) aircraft fleet is given in Chapter 9, along with descriptions of actual emissions databases. A subsonic fleet is included in all scenarios but is modified when an HSCT fleet is also present. The background atmosphere for the model scenarios was appropriate to either 2015 or 2050, though parametric studies of reactive chlorine (Cly) concentration and background SAD were also performed.

4.3.2.1. Background Atmospheres

Boundary conditions for the background atmosphere for long-lived gases for 1992, 2015, and 2050 (Table 4-8) are based on the IPCC IS92a scenario (IPCC, 1992, 1995). The boundary conditions used for halocarbons correspond to (a) 3.7, 3.0, and 2.0 ppbv of Cly in the upper stratosphere at 50 km at the Equator for 1992, 2015, and 2050, respectively; and (b) 15, 12.5, and 11.1 pptv of reactive bromine (Bry) in the upper stratosphere at 50 km at the Equator for 1992, 2015, and 2050, respectively.

Background sulfate SAD was supplied for the unperturbed stratospheric aerosol case as described in Table 8-8 in WMO (1992). This case is designated SA0 aerosols. To approximate the effects of an active period of volcanic eruptions (e.g., the past 2 decades), some scenarios also evaluate the effects of 4xSA0. Treatments of PSCs and associated chemical and dynamical processes differ among models.

4.3.2.2. Aircraft Emission Scenarios

The supersonic aircraft scenarios for 500 and 1,000 HSCTs are based on recent work (Baughcum and Henderson, 1998) carried out for the NASA technology concept aircraft HSCT, which would cruise supersonically in the 17-20 km altitude range. The NASA subsonic scenarios described in Chapter 9 account for displacement of subsonic air traffic by supersonic aircraft. Baseline computations assume supersonic aircraft emissions with EI(H2O)=1230 (1230 g H2O kg^-1 fuel), EI(NOx)=5 (5 g NO2 kg^-1 fuel), and a range of sulfate emission levels. Parametric studies were conducted around these baseline HSCT scenarios by investigating NOx emission index, fleet size, flight altitude, and background atmospheric conditions. These parametric studies are appropriate because the technology of a commercially viable supersonic airplane is not yet well-defined and results are needed to determine the sensitivity of the O3 impact to the technology level. A description of each of the scenarios evaluated by the participating models is given in Tables 4-4, 4-9, 4-10, 4-11, and 4-12. The HSCT scenarios use a number and letter designation preceded by the letter S (e.g., the first scenario is S1a). HSCT scenarios contain subsonic aircraft as well as HSCT commercial aircraft, with the combination accounting for the same passenger demand as in subsonic-only scenarios for 2015 and 2050. The HSCT scenarios in Tables 4-11 and 4-12 are generally analyzed relative to the corresponding 2015 and 2050 base plus subsonic scenarios.

The effects of sulfur, NOx, H2O, CO, and NMHC (as CH4) are simulated in the scenarios. The treatment of sulfur emission is discussed later in this section. For the other species, the aircraft effluents are put into the model as follows: Gridded fuel burn data (kg fuel/day) are first mapped into the model grid; the amount of material emitted into each grid box is given by the product of the fuel burn and the EI; and the emitted material is put into the grid box at each time step with the equivalent rate.

In this approach, we ignore the effect of plume processing and assume that these emitted materials are instantaneously mixed into the grid box. This assumption is probably valid in most of the stratosphere, though it may not be valid in the cold polar lower stratosphere during winter. In these regions, chemical processes are strongly nonlinear, thus raising concerns about the assumption. To date there have been no detailed wake model calculations supporting or rejecting this assumption. This important caveat should be remembered when considering the model results presented in this section.

The effect of sulfur emission on SAD depends on how much SO2 is converted to small particles in the aircraft plume. Conversion of SO2 to aerosol particles within the aircraft plume was found to perturb stratospheric SAD to a much greater extent than equivalent SO2 emissions (Weisenstein et al., 1996, 1998). Atmospheric measurements in the plume of the Concorde (Fahey et al., 1995) and other aircraft, along with modeling studies (Danilin et al., 1997; K�rcher and Fahey, 1997; Yu and Turco, 1998), suggest a minimum 10% conversion, with much larger conversion rates possible. However, near-field particle observations (as discussed in Chapters 3 and 7) suggest a typical conversion of sulfur to sulfuric acid (H2SO4) of less than 10% with a large, yet not fully understood variability under the assumption that the aerosol composition is H2SO4/H2O. Model sensitivity studies were designed to examine the full range of this uncertainty with conversion fractions of 0, 10, 50, and 100%.

In most scenarios, the reference atmosphere is consistent with a SAD using SA0. If the HSCT scenario assumes no sulfur emissions, SA0 is used for the HSCT case as well. In some HSCT calculations, the effect of increased SAD from SO2 emissions by aircraft is also considered, using a range of different sulfate conversion fractions (Tables 4-9 through 4-12). The additional SAD fields were constructed by calculating sulfate surface area for background and HSCT conditions using the AER microphysical model coupled to their 2-D CTM (Weisenstein et al., 1997). The SAD perturbation (HSCT-background) was derived in absolute units and was added to the SA0 background SAD distribution. The SA5 case assumes a 10% conversion rate for 500 HSCTs with an EI(SO2) of 0.4. This case also roughly approximates a 5% conversion rate, for an EI(SO2)=0.8, with the sulfur content of the fuel maintaining its current value. Figure 3-25 shows the annually zonally averaged perturbation of SAD (in �m² cm^-3) used in the SA5 scenario. SA6 is similar to SA5 but is based on 1,000 HSCTs. A 0% conversion with EI(SO2)=0.4 is considered in SA7 for an aircraft fleet of 500 planes. The assumption of 50% conversion with EI(SO2)=0.4 is considered here in SA1 and SA2 for aircraft fleets of 500 and 1,000 planes, respectively. The assumption of 100% conversion with EI(SO2)=0.4 is considered in SA3 and SA4 as an upper limit. SA3 assumes 500 aircraft, whereas SA4 assumes 1,000 aircraft. SAD was also computed for fleets flying 2 km lower. A summary of all SAD distributions used in these scenarios is given in Table 4-9.