dimensional model predictions of ozone response to ... - NASA

JOURNAL OF GEOPHYSICAL

RESEARCH, VOL. 100, NO. D2, PAGES 3075-3090, FEBRUARY

20, 1995

Sensitivity of two-dimensional model predictions of ozone response to stratospheric aircraft: An update David

B. Considine

Applied Research Corporation, Landover, Maryland

Anne R. Douglassand Charles H. Jackman NASA Goddard Space Flight Center, Greenbelt, Maryland

Abstract. The Goddard Space Flight Center two-dimensionalmodel of stratosphericphotochemistryand dynamics has been used to calculate the O• response

to stratospheric aircraft (high-speedcivil transport(HSCT)) emissions.The sensitivity of the model O• responsewas examined for systematicvariations of five parametersand two reaction rates over a wide range, expanding on calculations by variousmodelinggroupsfor the NASA High SpeedResearchProgram and the World MeteorologicalOrganization. In all, 448 model runs were required to test the effectsof variationsin the latitude, altitude, a,nd magnitude of the aircraft emissionsperturbation, the backgroundchlorine levels, the backgroundsulfate aerosolsurfacearea densities,and the rates of two key reactions. No deviation from previous conclusionsconcerningthe responseof O• to HSCTs was found in this more exhaustiveexploration of parameter space. Maximum O• depletionsoccur for high-altitude,low-latitudeHSCT perturbations.Small increasesin global total O• can occurfor low-altitude, high-latitude injections. Decreasingaerosolsurface area densitiesand backgroundchlorinelevelsincreasesthe sensitivityof model O• to the HSCT perturbations. The location of the aircraft emissionsis the most important determinant of the model response. Responseto the location of the HSCT emissionsis not changedqualitatively by changesin backgroundchlorine and aerosolloading. The responseis alsonot very sensitiveto changesin the rates of the reactions NO q- H02 • NO2 q- OH •nd HO2 q- 03 • OH q- 202 over the limits of their respectiveuncertainties. Finally, levels of lower stratosphericHO• generallydecreasewhen the HSCT perturbationis included,eventhoughthere are large increasesin H20 due to the perturbation.

Many recent studies have been conducted for the

Introduction

AESA (AtmosphericEffectsof Stratospheric Aircraft) componentof NASA's HSRP (High Speed Research Stratosphericaircraft (high-speedcivil transports Program)[Ko, 1992;Ko and Weisenstein, 1993;Ko and (HSCTs))arepotentialsources of stratospheric pollu- Douglass,1993]. Thesestudieshaveinvolveda number

tants. A fleet of such aircraft would directly inject NOx

of modeling groups. Each group reported their model's

(NOx = NO + NO2) andH20 into the stratosphere as prediction of Oa depletion to a few aircraft emissionsce-

well as other compoundssuch as unburnedhydrocar- narios. The scenarioswere carefully designedto be as bons,COe,CO, SOe,andsoot[Miake-Lye,1992].The realistic as possiblein the magnitude and distribution NO• and HeO injectionsfrom an economicallyviable of emissions,and correspondto different proposedfleet fleet may be largeenoughto significantlyaffectstrato- configurations.The studies then concentratedon comsphericOa levels[Douglassel at., 1992; Johnstonel paring the predictions of the different models to these

at., 1989].The possible response of stratospheric Oa

few realistic

aircraft

emission scenarios.

to a fleet of HSCTs has been studied extensively since One reason to limit the studies to a small number of the early 1970sbecauseof thesepotentially large effects scenarios is that it is economical to do so. The stan[Johnston,1971; Johnstonel at., 1989, and references dardization of the perturbations also aids in the intertherein]. comparison of the results of different models. In addi-

tion, if a model happens to be a completely accurate simulation of the real atmosphere, then the most realistic scenario will produce the best prediction of the atmosphericeffectsof stratosphericaircraft.

Copyright 1995 by the American Geophysical Union. Paper number 94JD02751.

0148-0227/95/94 JD-02751$05.00 3075

3076

CONSIDINE ET AL.: SENSITIVITY

Unfortunately, the validity of a model and its response to a perturbation is difficult to determine. As a result, it is hard to estimate the uncertainty of the model predictions. Intercomparison of the responses of different models to the same perturbation is one method. If there is a large intermodel differencethen the confidence in the model predictions will be low. If there is very little variation then the confidenceis higher, although the possibility does exist that all the models are incorrect in the same way. The scenariobased studies that

have been conducted

for the HSRP

use this method to evaluate the uncertainty of the model predictions. A secondmethod of determining the uncertainty is by a systematic study of the responseof a model to a wide range of perturbations and to a large variation in the input parameters of the model itself. If small changesin the perturbation produce large changesin model response,then the confidencein the model results will

be reduced.

Confidence

will also decrease if

OF MODEL OZONE RESPONSE

backgroundaerosolsurfacearea density are examined. In addition, the rates of two reactions, NO + HO2 --• NO2 + OH and HO2 + Oa --. OH + 202, are varied over their uncertainty limits with fixed backgroundchlorine and aerosolloading. These two reactions were chosen because the first is critical

to interference

between

the

NO• andHO• (HO• = H + OH + HO2) catalyticcycles that destroy O3, and the second has both a large uncertainty and an important role in determiningthe rate of HO•-induced O3 loss.

The organization of this paper is as follows: First, the GSFC two-dimensionalmodel is described,including the different backgroundaerosoland chlorineassumptions that are varied in this study. Second,the aircraft fleet perturbations are presented. Third, the response of global total ozone and column ozone to variations in fleet location and magnitude are examined. Following this, the responseto changesin backgroundchlorine and aerosolloading is presented. Finally, the changes induced by reaction rate variations are discussed.

the variation of a model parameter (suchas a reaction rate or boundarycondition)produceslarge changesin the model response.However, if large variations in the Model Description perturbation or the input parameters do not have a The GSFC two-dimensional stratospheric photolarge effect on the model predictions, then the confidence that the model predictions are correct will be chemistrymodel [Douglasset al., 1989; Jackmanet al., 1990]has a 10ø latitudinal resolution,extending increased. In a recent evaluation of NASA's HSRP, the National Research Council's Panel on Atmospheric from-85 ø to +85 ø. It has 30 levels equally spaced in

logpressure, fromthe groundto 0.23mbar(• 60 km in approximately 2-km increments.)The modeltime step studiesof this type be conducted[Graedelet al., 1994].

Effects of Stratospheric Aircraft suggestedthat more

This paper details the resultsof a study of this second type, where the responseof a model to systematic variations in both perturbation and input parameters is explored. Its primary goal is to evaluate the uncertainty of the model prediction of Oa responsethrough parameter variation. A secondarygoal is to examine the mechanisms that produce changesin model responsewhen the HSCT perturbation or the model is changed. Other papers have dealt with the mechanismsin more detail. For instance,Considineet al. [1994],Tie et al. [199q], and Pitari et al. [1993]examinedthe effectsof polar stratospheric clouds(PSCs)on HSCT perturbationpredictions;Bekki and Pyle [1993]and Weisensteinet al. [1993]studiedthe impact of sulfateaerosols,and Jack-

is I day. A residual circulationis calculatedusing a 4-yearzonalaverageof NationalMeteorological Center temperatures. Heating rates are taken from Dopplick et al. [1974,1979]from the groundto 100 mbar,and from Rosenfieldet al. [1987]from 100 mbar to the top of the model. Family chemistryapproximationsare

by exclusivelyusing the Goddard Space Flight Center

the aircraft and lossdue to rainout and photochemistry. The model constituent distributions coinparefavorably

used. The model calculatesdaytime averagevaluesof 55

species.Of these,25 are transported. Brominechemical reactions are included as well as the C10 dimer

reactionsimportant to the polar regions. The model H20 distribution consists of two components. First,

a backgroundH20 distributionderivedh'om limb infrared monitor of the stratosphere(LIMS) data is included,as describedby Jackmanet al. [1990]. This man et al. [1991]lookedat the effectsof changes in the distributionis changedmonthly. Second,the H20 injected by the subsonicand supersonicaircraft fleetsis model residual circulation. This work is similar to that of Johnstonet al. [1989]. It differsfrom that study a transportedquantity with a productionterm due to

(GSFC) two-dimensional model[Douglass et al., 1989; Jackmaneta!., 1990] instead of being basedprimarily on one-dimensionalmodel results. In addition, the model incorporatesheterogeneousreactionson a strato-

sphericsulfateaerosol(SSA) layer;the Johnstonet al. study consideredonly gas phase reactions. Heterogeneous chemistry has since been found to be critically important to the chemistry of the lower stratosphere and in particular to its responseto NOx perturbations

with those obtained from other two-dimensional

models

[Jackman et al., 1989;PratherandRemsberg, 1993]. Three heterogeneousreactionsare included in the model formulation:

C1ONO2+ H20 C1ONO2+HC1 N20• + H20

--* HNOa +HOC1 --• HNOa +C12 --• 2HNO3.

(1) (2) (3)

[Weisensteine! al., 1991]. Heterogeneous reactionson PSC surfaces are not considered.

Model responseto variations in the altitude, latitude and magnitude of injection, background chlorine, and

These reactions occur in the atmospherevia interactions with aerosolparticle surfaces. The surfaceinteractions result in reaction rates far higher than would

CONSIDINE

ET AL.: SENSITIVITY

OF MODEL

OZONE

RESPONSE

3077

occurby a purelygasphasepathway[Hansonand Rav- Table 1. Chlorine BoundaryConditionsfor the Low, Medium,and High ChlorineConditionsUsedin This ishankara,1991]. Heterogeneous reactions are implemented in the Study model by specifyinga sulfate aerosolsurfacearea denSpecies sity A at each model grid point. The surface area density is the total surface area available on the aerosol CFC-11 CFC-12 particles contained in a unit atmospheric volume. A CFC-113 "sticking" coefficient,7, is also specifiedfor each reCFC-114 action. The sticking coefficientcan be interpreted as CFC-115 the probability that a collisionof a reacting molecule CC14

HighCly

MediumCly

LowCly

260 510 70 10 8

124 359 49 7.8 7.2

50 233 31 6.2 6.6

(C1ONO2or N205) with the aerosolparticlesurfacewill

100

34

12

HCFC-22

200

3.7

0.15

result in a heterogeneousreaction. The total heterogeneousreaction rate alsodependson the thermal velocity of the incident molecule, v. Thus the total reaction rate is given by:

CH3CC13

150

0

0

2

0.2

0

600

600

600

Halon 1211

CH3C1

ColumnI givesthe constituent,and columns2, 3, and 4 give the groundmixingratioboundary condition for that A(ch, z) (4) constituent, in parts per trillion by volume(pptv). The highandmediumcases arethesameaswasusedin thethird In this equation, j - {1,2,3} denotesthe number Atmospheric Effectsof Stratospheric Aircraftreportwritten of the heterogeneousreaction listed above, and I - for the High SpeedResearchProgram[Ko and Douglass, {C1ONO2,C1ONO2,N205} indicatesthe moleculeinci- 1993]. dent on the aerosolparticle [Hofmann and Solomon, 1989]. The thermal velocity of the C1ONO2 and N205 moleculesare calculated using the formula vl =

used was not calculated for each type of molecule as a

V/(8k'T/•rMl), wherek' is Boltzmann 's constantand function of temperature. Instead, it was taken to be

Ml is the molecular weight of the incident molecule in the reaction.

a constantcorresponding to a moleculewith a molecu-

lar weightof 100at a temperature of 204.4K [Ko and

The stickingcoefficient usedfor reaction(3) is a con- Weisenstein,1993]. Reaction(2) was not includedin reactionset, and the stickingcocfstant,7a = 0.1. Reactions(1) and(2) aresensitivefunc- the heterogeneous ficient for reaction (1) was taken to be a functionof tions of the weight percent of H2SOq characterizingthe SSA [Hansonand Ravishankara,1991]. Denotingthis the zonalmeantemperature,7[ -0.006exp[-0.15(T-

200)]. Runstestingthe modelsensitivityto the same

value by W, then

71 -- 10(1'87-0'074W)

72 _ 0.171.

(5)

The weight percent of H2SO4 depends on the H20 partial pressure and on the zonal mean temperature. The relationship used is taken from data published by

Steeleand Hamill, [1981]. Four different assumptionsfor the SSA surface area density distribution are used' a no aerosol case and three others.

The

basic distribution

A is the same in

each of these cases. This is multiplied by a prefactor f, where f = 1, 2, or 4 to give a low, medium, and high surface area density, respectively. The basic distribution is taken from World Meteorological Organization

(WMO) [1992]and is basedon measurements takenby the SAGE

II satellite.

This base distribution

was also

HSCT perturbation have been made using both prescriptions,with only smalldifferences in the results. The three sets of boundary conditionsemployed in this study correspondto low, medium, and high levels of backgroundchlorine. Other sourcegas concentrations correspondto those expectedfor the 2015 atmosphere.The high and mediumbackgroundchlorine conditions are those used in the scenario calculations,

and contain3.7 parts per billion by volume(ppbv) and 2.0 ppbvchlorine,respectively. The lowchlorineboundary conditionshavea total chlorinecontentof 1.4 ppbv. The valuesof the differentchlorinesourcegasesfor this boundaryconditionwereobtainedby assuming that all CFCs would be phasedout in 1995, and concentrations of the gaseswoulddecayfor 100yearsaccordingto the lifetimesgivenin chapter8 of WMO [1992].The values of the different chlorinecompoundsusedin the bound-

usedin the HSRP scenariocalculations. The f - 1 case representsbackgroundsulfate conditionsseveral years ary conditionsare given in Table 1. after any large injectionsof sulfur into the stratosphere.

The high distribution(f = 4) corresponds to the me- Aircraft Fleet Assumptions dian value of the SSA surface area densities over the

last two decades[WMO, 1992]. Althoughthis can be thought of as a volcanicallyperturbed sulfate distribution, the surface area densitiesare far smaller than can occur immediately after a major volcanic eruption. This heterogeneous chemistryprescriptionis roughly comparable to the one used in recent scenario calculations

conducted

for the HSRP.

There

are some dif-

ferences,however. In those runs, the thermal velocity

Emissions from a subsonic aircraft fleet are included

in all of the simulationsreported. The subsonicfleet is assumedto emit only NOx and H20, as given by Ko

[1992].The magnitudeof the aircraftemissions canbe characterizedby the total fuel consumedper year by the fleet, and an emissionindex (EI), whichis the number of gramsof pollutant releasedinto the atmosphereper kilogramof fuel burned.The magnitudeof the subsonic

3078

CONSIDINE

ET AL.: SENSITIVITY

emission is 170x 109kg/yr, with an EI of 20.7for NO•

OF MODEL

OZONE

RESPONSE

Table 2. Parameters Varied in the Sensitivity Runs

and 1230 for H20 emissions. The latitudinal distribu-

tion peaks in the northern midlatitudesand tapers off to the north

and south.

The annualHSCT fleetfueluseis 70 x 109kg. Three differentEIs for NO• of 5, 15, or 45 are usedto specify a low, medium,and high amountof NOs injected by the HSCTs into the stratosphere. In all casesthe EI for H20 injected by the stratosphericaircraft fleet is 1230. The HSCT

emissions are distributed over six

Parameter

Low

Medium

High

EmissionIndex

5 g/kg

15 g/kg

45 g/kg

Latitude

0ø

30øN

60øN

Altitude

120 mbar

68 mbar

38.5 mbar

Chlorine

1.4 ppbv

2.0 ppbv

3.7 ppbv

Aerosol

A

2A

4A

k•o,Ho• kHo•,oa

k- a k - er

k k

k+ a k+ a

adjoiningmodelgrid boxes,two in latitude and three in altitude. Half the material is injected at eachof the two

Runs were made for every possible combination of the first five parameters. In addition, a zero background aerosol latitudes.Of the materialinjectedat a singlelatitude, casewas included and runs were made for every combination half is injectedat the centralgrid box and the remain- of emissionindex, fleet location, and chlorine background. Runs made in which the final two parameters were varied ing half is dividedbetweenthe upperandthe lowergrid used the medium values of background chlorine and aerosol boxes. The fleet emissions are therefore distributed over loading. Runs were then made with every combination of a region spanning20ø latitude and about 6 km in alti- fleet location and magnitude of emissions.The low and high tude. This spatial distributionof emissions is designed uncertainty limits for each of these reactions was determined

to allow easy testing and interpretation of the model responseto changesin the fleet location over a large range of altitudes and latitudes. It doesnot conform to

usingvaluesobtainedfrom DeMore et al. [1992].

expected latitudinal or vertical emission distributions baseonly by the addition of the HSCT perturbation. In based on current flight corridors. termsof A GO• sc*, the runsobeythe following genThe center of the distribution specifiedin the previeral relationships: (1) GOa HscTdecreases as the fleet ous paragraph is varied over three model pressurelevels altitudeis increased (negative change in AGO•SCT). to examine the sensitivity of the model responseto the (2) GOa HsCTincreases asthefleetlatitudeis increased location of the aircraft emissions. These correspondap-

(positive change in AGOa•SCT). (3) Thereis a positive

proximatelyto 15, 19, and 23 km in altitude (120 mbar, change in AGOa •scT forincreased levels ofbackground 68 mbar, and 38.5 mbar). The distributionis alsovar- chlorine. (4) Thereis a positive change in AGO•scT ied over three latitudes (0øN, 30øN, and 60øN), for a for increasedlevelsof backgroundaerosols. total of nine different fleet locations. Note that

in cases 3 and 4 we are interested

in the

effect that changesin backgroundchlorineor aerosol Results surface area density have on the model responseto the HSCT perturbation. This should not be confusedwith The study consistedof the systematicvariation of the direct responseof the model to changesin backbackgroundchlorineover the low, medium,and high ground chlorineor aerosols.There are significantvarichlorine backgroundsand the aerosolsover the zero, ationsin GO•^s•:fromcaseto case. low, medium,and high casesfor the 28 possiblecombiThe response of AGOa•scTto increases in emission nationsof HSCT fleet locationand magnitude(includ- index is more complicated and depends on the fleet ing onebaserun). This portionof the studyrequired location as well as background amounts o! chlorine 12 x 28 or 336 runs. In addition, the two reaction rates

were varied over their low and high uncertainty limits for the medium chlorine and aerosolcase. This required an extra 4 x 28 or 112 runs, for a total of 448 model runs. The parametersvaried in this study are shown

and aerosols. When the fleet location is at high altitudes and low latitudes, increases in emissions will

produce a strong negative response. However, in the

low-altitude, high-latitude runsA GO•scTcanincrease

with increasingaircraft emissionsdue to lower stratoin Table 2. All of the base runs included the subsonic sphericHO•- NO• interactionsand NOs participation emissionsspecifiedabove, so the changesbetweenthe in CH4 oxidation. These changeswill be studiedin more baseand the perturbed runs are due to the HSCT emis- detail below.

sionsonly. Representativeresultsfrom theseruns will

Effects of Changes in Fleet Location The responseof the model to the HSCT fleet can The basicbehavior of AGOa •scT canbeseenin Figbe quantifiedby the changein globally and annually ures la, lb, and lc. Figure la showsthe variation in averagedcolumnOa betweenthe perturbedcaseand

be discussedand presented below.

the base case:

AGOa •sc* asa functionof emission indexforthethree

pressurelevels of injection. The injection latitude is 0ø. Figures lb and lc show the same results for the 30ø

/XGo•sc• - GOa "sc•' - Go•^Se and 60ø injection latitudes, respectively.All of the fig_ oo^se . (6) ures are for the middle backgroundchlorineand aerosol Here,GOa •^sE is the globallyandannuallyaveragedsurface area densities. Examination of these plots shows that the response columnozonewithoutthe HSCTfleet,andGO•scTis emission indexis a sensithe samequantity for a run which is changedfi'om the of AGO• scT to increasing

CONSIDINE ET AL.. SENSITIVITY OF MODEL OZONE RESPONSE

3079

rive function of fleet location, and in particular, fleet altitude. The largest depletion for these chlorine and aerosolamountsof about -7.5% occursfor a fleet posi-

-2

tion of 0ø, 38.5 mbar, and E1 = 45. At low altitudes

andhighlatitudes,AGOa HsCTis positiveandincreases

-4

with increasingemissionindex, as shownin Figure l c.

Here,AGOa HsCTis +0.25%for the 60ø, 120mbar,EI

-6

= 45 case. The largest sensitivity to emissionindex oc120

-8

.....

cursfor highaltitudefleetlocations.Increasingthe fleet latitudereducesthe sensitivityof the modelto changes in the emissionindex. At high altitudesand latitudes,

MB

68bib 38.5

-10

bib

I

I

10

20 EMISSION

I

thereis an increase in the sensitivity of AGOa HsCTto

I

30

40

5O

INDEX

emissionindex at higheremissionrates. This is seenby the steeper slope between the EI = 15 and the EI = 45 cases,as comparedto the slope betweenthe EI = ,5 and the EI = 15 casesin the Figure lc 38.5-mbarplot. Finally, for low-altitude fleet emissionsthe model is rel-

atively insensitiveto increasesin NOs at all injection latitudes.

Changesin backgroundchlorineand aerosolsurface areadensityaffectthe behaviorshownin Figure1 quantitatively but not qualitatively. The effectswill be discussed in more detail below.

The responseof 03 to variations in the location and 120

....

magnitude of the HSCT NO•. emissionsdependson three factors: (1) the HSCT-inducedincreasein a.tmo-

MB

68 MB 38.5 I

"',,.

•..

MB I

10

2o EMISSION

'.,, I

.

30

I

4o

50

sphericNOu load!ng(whereNOy is the sumof all odd nitrogenspecies: NOy = N+NO+NOe+NOa+2N2Os+ HOeNOe+C1ONOe+HNOa); (2) thepartitioningof the NOu increase betweenactiveandinactiveNOu;and(3) the spatial distribution of the increase. Table 3 shows

INDEX

howthesefactorscombineto producethe Oa response to the HSCT

fleet.

The first three columns list the

magnitude, latitude, and approximate altitude of the perturbation. Column 4 lists the annuallyaveragedin0

creasein the amountof atmospheric NOu, expressed in moleculesof NOy. Column5 showsthe increasein

NOs;column6 givesthe active/inactive NOu ratiofor o

,o -1

the increase;column 7 givesthe changein the num-

ber of Oa molecules; and column8 givesthe number



•

••

8----

0.•

sphereOa decreaseseenin Figure 3a. Maximran in- -- -30 creasesin the northernhemisphereof about 20% are lo- •

three-dimensional modelcirculationexhibitedmorevigoroushorizontal and vertical tracer transport. Clearly,

10a_

,

,

-60

-30

,

,

,

0

30

60

0

90

_

LATITUDE

and middle valuesof chlorineand backgroundaerosols tions as Figure 4a.

L_----';i '. 100

C21

-

20> o

o

10a_

0

1000

-60

-30

0

30

60

90

LATITUDE

Figure5a showsthe changes in O3 in September, for the low-altitude,high-latitudefleet. The O3 lossesare factorof 30 smallerthan in the previouscase(Figure but have the same spatial distribution. The northern hemisphere troposphereand lowerstratosphere is characterizedby 03 increasesof up to 5%. At 60øN,O3 increasesoccurup to about 50 mbar. In the low-latitude,

60

•

b

50 L,.j 40 D 30

high-altitudecasethe increases turn to losses at sig-

theNOy increase doesnotshowa double-peaked distri-

•0

....

I: i.o

-90

nificantlylower altitudes, between100 and 200 mbar. Figure5b showsthe NO• increases for Figure5a. The plot showstwo peaksin the NO• increase.Note that

0.1 .__

40 D

Figure 4. (a) Percent changein O3 fi'oln the aircraft perturbation as a function of latitude and pressure, for September. The fleet is at the low-lati[ude, high-altitudelocation(0ø, 38.5 mbar). The EI = 45

wereused.(b) Changein NO• (NO• = NO + NO2), in partsper billion by volume(ppbv), for the samecondi-

-- ......

100

--

'

o

_

_

10oo

10 a... 0

bution. It peaksat the locationof the HSCT fleet and -60 -30 0 30 613 90 -90 decreases away from it. The NO• increaseis doubleLATITUDE peakedbecausethe NO•/NOy ratio increases toward the equator,so an inceasingfraction of the increasein Figure 5. (a) Percent changein O3 from the airNO• is in the form of NO•. This resultsin an equa- craft perturbation as a function of latitude and prestorial increase in NO• of about 0.3 ppbv at 10 mbar, sure, for September. The fleet is at the high-latitude, over 40 times smaller than the peak increasesin NO• low-altitudelocation(60øN, 120 mbar). The EI = 45, seenin Figure 4b. The approximately0.6 ppbv lower and middle valuesof chlorine and backgroundaerosols stratosphere/uppertropospherepeak in the northern wereused.(b) Changein NO, (NO• = NO + NO2), in hemispherehigh latitudesseenin Figure5b is about 6 ppbv, for the same conditionsas Figure 5a.

CONSIDINE

ET AL.' SENSITIVITY

OF MODEL

OZONE

RESPONSE

3083

if the two-dimensionalmodel had more vigoroushorizontal and vertical transport, there would be lessisolation of the Oa lossregion mentionedabove to the emissionsof a high-latitude fleet. In such a situation there would be less of a differencein impact on Oa between fleets with

different

locations.

The results of this study reinforce the observation

that transport processes may be a large sourceof model error. They also suggestthat not all fleet dispositions are equal. A fuel efficient fleet with low emissionsin one location may be a source of more O• loss than a more polluting one that flies at higher latitudes and lower

lO0O

-90

altitudes.

-60

-3o

o

3o

60

90

LATITUDE

Effects of Changes in Fleet Location on H_,O and HOx

Changes in fleet location also have a large impact on the HSCT-induced increase in Ho_O. Figures 6a and 6b show the September increase in HoO caused by the HSCT fleet for the low-latitude, high-altitude and high-latitude, low-altitude cases,respectively.The maximum increasein H20 for the low-latitude, highaltitude injection is about 1.5 parts per million by vol-

••o(•• '3øø' .oof• 60•

1

-'-'

50•

10

•'øø'/:-30•' -•_ -

.-'"t'5-? ...... lOO

"

'-

I..,_ _

ø

0 _ _

lOOO

-90

,

-60

-30

,

,

,

-0

0

30

60

90

LATITUDE

1

l•igure 7. (a) Septemberchangesin HO• (HO• = H + OH + HO•) from the aircraft emissions,in parts per trillion by volume(pptv). The fleet is at the lowlatitude, high-altitudelocation(0ø, 38.5 mbar), and has an EI = 45. (b) Sameas Figure 7a, for EI = 5.

lO

lOO

lOOO

ume(ppmv), occurringat the equatorat about40 mbar.

-90

-60

-50

0

50

60

90

LATITUDE !

i

i

i

b

-6o

1-

_•o.o• •

•

-50

o/••___= 40

10 = -

---50

Fairly large increasesin water vapor occur throughout the stratospherewith a band of over 0.8 ppmv increases extending from pole to pole. The increasesare fairly symmetrically distributed between the northern and southern hemispheres. In contrast, the high-latitude, low-altitude injection location produces a smaller and much more unevenly distributed H•O increase. Southern hemisphereincreasesdo not exceed0.03 ppmv. The northern hemispheredistribution peaks at 60ø and 120

mbar (the centrallatitude and pressurelevelof the fleet location). Maximumincreases of about 0.4 ppmvoccur here but are confined to the lower stratosphere.

100

10

1000 -90

,

lorn

-60

-30

0

',

,

30

60

0 90

LATITUDE

t;'igure 6. (a) Septemberincreasein It20 fi'omthe aircraft emissions, in parts per million by volume(ppmv). The fleet is at the low-latitude, high-altitude location

The changesin HOx from the HSCT perturbations are complicated and depend sensitivelyon the amount of NOs depositedby the fleet. Figures 7a and 7b show the change in HO• for the low-latitude, high-altitude fleet location. Fleet emissionsat this location produce the largestincreasesin H20 vapor. Figure 7a showsthe

changein parts per trillion by volume (pptv) for the EI = 45 case, while Figure 7b showsthe HO• change

(0ø, 38.5 mbar). (b) Sameas Figure6a, with the fleet for the EI = 5 case. The equatorial midstratosphere is at the high-latitude, low-altitude location (60øN, 120 characterized by HO• decreasesof up to 20 pptv in the mbar). EI = 45 caseand up to i pptv in the EI = 5 case. HO•

3084

CONSIDINE

ET AL.'

SENSITIVITY

OF MODEL

decreasesin the lower stratosphere even though H20 increasesbecause HO• production also dependson O• perturbation. In the upper stratosphere where the O• decrease from the HSCT NO• is small, the H20 fi'om The increase is similar

RESPONSE

.... LOW CHLORINE"•", '"' ß MEDIUM /) ,]

(O• = Oa+ O + O(• D)), whichisreduced by theHSCT the HSCT fleet increases HO•.

OZONE

10

in both cases.The partitioning of HO• dependson both O• and NO•, and this will also affect the responseof HO• to the HSCTs. This lower stratospheric decrease in HO• reduces the loss of Oa due HO• catalysis. The larger the increase in NO•, the more pronouncedthe

1GO

reduction in HO• and HO• loss.

The response of HO• to the HSCTs indicates that the large increasesin H20 producedby the HSCTs do not have a large impact on Ga. It shouldbe noted that this conclusion

does not consider

1

lO

1GO

lOOO

MIXINGRATIO(PPTV)

the effects of PSCs.

However, Pitavi et al. [1993] and Considineet al. [1994]have found that PSCs reducethe sensitivityof

LOW AEROSOLS

the model to HSCT perturbations slightly, so it is likely

MEDIUM

that the effects of H20 emissions on Oa are small even when PSCs are considered. 10

Effects of Changes in Aerosol and Chlorine The rates of the heterogeneousreactions (reac-

tions (1), (2), and (3)) are increasedby increases in the SSA surface area density. These reactionshave two basic effects. First, they convert active odd nitrogen to reservoir forms. Second, reservoir chlorine is converted to active chlorine. The chlorine repa.rtitioning happenseither by direct conversionof inactiveto active

chlorinewith reactions(1) and (2) or indirectlyfi'oln the odd nitrogenrepartitioning,as with reaction(3). The repartitioningreducesthe fractionof total O3 loss that is causedby NOx speciesand increasesthe role of

1GO

1

lO

1GO

lOOO

MIXINGRATIO(PPTV) Figure 8. (a) Increasein lowerstratosphericactive chlorine due to increasesin backgroundchlorine levels.

Shownis Clx (= C1+ C10) as a functionof pressure

at 55øN in September. The low chlorine case correspondsto 1.4 ppbv tota,1chlorine, the medium casecorhave less of an impact on O3. A model with higher respondsto 2.0 ppbv, and the high case to 3.7 ppbv

odd chlorine. Increases in NOx under these conditions

heterogeneous reaction rates will be less sensitiveto HSCT perturbations; as aerosolsurfacearea densities decrease,modelsensitivitywill increase[Weiscnsteine! al., 1993]. Decreasesin backgroundchlorinelevelsshouldalso increasethe impact of the HSCT perturbation. The mechanismis essentially the same as describedabove. Lower amounts of ba.ckgroundchlorinesuch as might occur late in the next century will result in less of

background chlorine.(b) Increasein lowerstratospheric active chlorine due to increasesin background aerosol levels. The location and time is the same as Figure 8a The low aerosol case correspondsto f = 1, medium correspondsto f = 2, and the high aerosolcasecorresponds to f - 4. Here, f is the prefactor applied to

the basicstratosphericsulfateaerosol(SSA) distribution described

in the text.

the HSCT perturbation being convertedto inactive C1ONO2 so NO• increaseswill be larger. Also, the relFigure 9a showsthe responseof global total Oa to ative importance of chlorine loss processesin the Oa increasesin the aerosoland chlorine loading of the atbudgetis smaller,sothe role of NO• will be larger,also mosphere.This plot is for a fleet located at low latitudes increasingthe modelsensitivityto HSCTs [Johnstone! (0ø) andhighaltitudes(38.5mbar),with an emission inal., 1989; Weiscnstein½! al., 1991; I(o and Weiscn- dex of 45. From this plot it is easy to seethat increasing stein, 1992; Ko and Douglass,1993]. aerosolloadingreducesthe O• depletionfor all chlorine Thus the impact of changes in aerosol and back- levels.The largestsensitivity of AGOa HscTto aerosols groundchlorinelevelsis relatedto their effecton lower occurs forlowaerosol amounts, withAGOa HscTreduced stratospheric Cl• (Cl• = C1+ C10). Figures8a and 8b by a factor of about 1/3 as the aerosolbackgroundinshowthe changein lowerstratosphericCl• for variations creasesfrom no aerosolsto the f = 1 case. The response HscTto furtherincreases in aerosols fiattens out in backgroundchlorineand aerosollevels,respectively. ofAGOa The plot is for Septemberat 55øN. A factor of 3 increase for higher aerosolamounts. The figure also showsthat in Cl• occursfor the variation in backgroundchlorine, increasingbackgroundchlorine also reducesOa deplewhile a factor of 4 increasein backgroundaerosolsin- tion.Thereduction in AGO•scTresulting fromanincreasesClx by about a factor of 2 at midlatitudes.

creasein aerosolsfrom f = i to f = 4 is larger than the

CONSIDINE ET AL.' SENSITIVITY -4

3085

Figures 8a and 8b. Increasesin backgroundchlorine do not produce as much of a reduction in NO, as aerosol increases do. Because the aerosols both reduce NOs and increase CI•, they can have a greater impact on

-6

--

I'"'

AGOa nscTeventhoughtheincreases in CI• aresmaller than occur by increasingthe chlorine background.

II'

/'/i t/7 ' -t2. .,";,•/ -14

OF MODEL OZONE RESPONSE

INE - LOW CHLOR .... MEDIUM

ii/

HIGH

-16

I

1

I

2

BACKGROUND I

I

3 AEROSOL

!

i

I

4 FACTOR i

Figure9aalsoshows thatthesensitivity ofAGO3 nscT to increasesin aerosolis slightly modified at higher chlo-

rinelevels.Forinstance, thedecrease in AGO3 nscTbetween the f = 2 and f = 4 casesis marginally greater for higher chlorine than the lower chlorinebackground. The effect is not very large, however. The same holds

true for the sensitivityof AGO3 nscT to increases in chlorine: there is a small increase in t.he responsewith changingincreasingaerosolloading.

Figures9b and 9c showthe response of AGOa nscT to changesin chlorine and aerosol loading for higherlatitude, lower-altitude fleet locations. In Figure 9b the fleet is at 30øN, 68 mbar and in Figure 9c the location is 60øN, 120 mbar. The qualitative features of Figure 9a are preserved although there are large changesin the

-1

I'I'",.•,• ,'

absolute valuesof AGO•scT and the variationfrom //•)' ' LOW CHLORINE //• .... MEDIUM

low to high levels of aerosoland chlorineloading. Column 8 of Table 3 lists the changein the number of -5 Oa moleculesin the atmosphereper additional molecule of NOs added by the stratospheric aircraft. The table -7 shows that the response of Oa depends sensitively on the location of the aircraft fleet. Figures 10a-10d show 1 2 3 4 how this relationship changesas chlorine and aerosol BACKGROUND AEROSOL FACTOR levels in the model change. Figure 10a showsthe rela1.0 tionship for the EI = 5, low backgroundchlorine, low backgroundaerosol condition. At the highest injection altitudes the responseof the model is not very sensitive 0.5 to changesin fleet latitude. At the lowest altitudes the sensitivity to changesin latitudes is a maximum. Figure 10b showsthe relationship for the high background ,!/, 0.0 chlorine, low aerosol case. For high-altitude injections / INE there has been a significant decreasein model sensitiv///' .... MEDIUM ity. The changeis somewhatlessfor the low-altitude in-0.5 jection. Figure 10c showsthe low chlorine, high aerosol case. This figure is very similar to Figure 10a, showing -1.0 I ! I I that aerosoleffectsare most important in high chlorine environments. Figure 10d showsthe combinedeffectsof 0 1 2 5 4 5 both high chlorine and high aerosolbackgrounds.This BACKGROUND AEROSOL FACTOR case exhibits the largest range of fleet locations which Figure 9. (a) Changein globallyand anually av- produce increasesin Oa from the aircraft fleet. At low eragedcolumnozone(AGO3 •scT) with backgroundfleet altitudes, however,the changein the HSCT impact

/

...... HIGH

I

I

I

!

i

!

I

............

,•/

......HIGH

aerosols andchlorine loading.AGO3 nscTis the per-

cent changewhen the HSCT fleet emissionsoccur at the low-latitude,high-altitudelocation(0ø, 38.5 mbar), for an emissionindex of 45. The background aerosol factor is applied to the basicsulfate aerosoldistribution to adjustits magnitude.(b) Sameas Figure9a, for a middle-latitude,middle-altitudefleet location(30øN,

on Oa varies with fleet latitude about as much in this

case as in the other three. At high injection altitudes the variation with fleet latitude is somewhat greater.

The largesensitivityof AGO3 •scT at low aerosol loading(f = 0 to f = 1) indicatesuncertaintyin the

response of AGO3 nscTduringvolcanically quiescent pe-

68 mbar). (c) SameasFigure9a, for the high-latitude, riods. This is especiallyso if the true SSA background is smaller than the f = 1 case that is assumed to replow-altitudefleet location(60øN,120mbar). resent background conditions in this study.

al. reduction causedby the increasein backgroundchlorine levels. This is interesting becauselower stratospheric active chlorineincreasesmore fi'om the changein backgroundchlorinethan the changein aerosols,as shownin

Toon et

[1979]concludedthat the sulfate aerosolat that

time could be accounted for by observed concentrations

of tropospheric OCS and SO2 This suggeststhat the f = 1 case does represent a volcanically unperturbed situation

and minimal

SSA surface

area densities.

A

3086

CONSIDINE

22

SENSITIVITY

a ................

---'

• 19.,..,,-,'""

*-'""'

/'"'

' ..--

.-'" ..-'

.,,.,,

/,•

,•P -"

.,,.,'

.,,

,. .. ,½'o..' •.-

RESPONSE

21

.--' /.,,

OZONE

22 _•

20

.