simulated global distribution and deposition of reactive ... · tel/us (1989),4/8,256-271 simulated...

Tel/us (1989),4/8,256-271

Simulated global distribution and deposition of reactivenitrogen emitted by fossil fuel combustion

By HIRAM LEVY II and WALTER J. MOXIM. Geophysical Fluid Dynamics Laboratory/NOAA,Princeton University, P.O. Bo.~ 308, Princeton, NJ 08542, USA

(Manuscript received 12 April 1988; in final form II October 1988)

ABSTRACTWe use the medium resolution (- 265 km horizontal grid) GFDL general circulation transportmodel to simulate the global spread and deposition of reactive nitrogen emitted by fossil fuelcombustion. The nitrogen species are transported as a single tracer with no explicit chemistry.Chemical reactions are only present implicitly in the bulk coefficients for dry and wetremoval. The observed wet deposition of nitrogen over North America is used to determinethe global parameter for wet deposition, and constant bulk coefficients for dry deposition overland and sea are pre-calculated from measured concentrations and deposition velocities. Thesimulated yearly depositions in Western Europe and at regional export sites, as well assimulated yearly concentrations and their seasonal variation over the North Pacific, arecompared with available observations. The agreement is generally quite good and almostalways within a factor of 2. This model is then used to identify a number of important sourceregions and long-range transport mechanisms: (I) Asian emissions supply two-thirds of thesoluble nitrogen compounds over the North Pacific. In the summer, North Americanemissions are important over the subtropical North Pacific. (2) Nitrogen emissions fromEurope dominate the nitrogen component of Arctic haze in the lower troposphere, whileNorth American and Asian emissions are only important locally. The model predicts a largegradient in the Arctic with average winter mixing ratios ranging from less than 0.1 ppbvoverAlaska to more than I ppbv over eastern Russia. (3) Throughout the Southern Hemisphere,the emissions from fossil fuel combustion account for 10% or less of the observed solublenitrogen at remote sites, an amount less than a previously simulated contribution fromstratospheric injection. The long-range transport of PAN, NO, production by lightning andbiomass burning, and some, as yet, unknown marine biogenic source may all supply part ofthis background soluble nitrogen. However, the similarity between the seasonal cyclesobserved at Samoa for soluble nitrogen and for 0), a species known to be supplied from thestratosphere, suggests a major role for either stratospheric injection or an upper troposphericsource.

t. Introduction

Reactive nitrogen compounds playa major rolein the atmospheric chemistry of the troposphereover a wide range of spatial scales andenvironmental issues. The impact of nitrogenoxides on photochemical smog, a local phenom-enon, has been well documented (e.g., Leighton,1961), and the regional problem of acid rain, ofwhich acidic nitrogen is one of the two principalcomponents, has also been widely discussed (e.g.,NRC, 1983). Through their impact on the level of

Tellus 418 (1989), 3

tropospheric 03 (Crutzen, 1974) and OH (Levy,1971), nitrogen oxides playa major role in theglobal reactivity of the troposphere. This reac-tivity, in turn, controls the level of a number ofgreenhouse gases that influence the environmenton a global scale. At this time there is insufficientobservational data. to determine the globaldistribution of reactive nitrogen. This globalsimulation provides a first look at suchdistributions.

The largest source of reactive nitrogen in theglobal 1itrogen cycle, the combustion of fossil

DEPOSITION OF REACTIVE NITROGEN EMITTED BY FOSSIL FUEL COMBUSTION 257

driven by 6-h time-average winds and precipi-tation that are generated for one year by theparent general circulation model. This parentGCM has no diurnal variation of insolation andwill not realistically simulate atmospheric fluctu-ations with periods shorter than 6 h. However, ithas been very successful in simulating manyfeatures of the earth's climate (Manabe et al.,1974; Manabe and Holloway, 1975), and resolvesweather events such as frontal passages, anti-cyclones, and mid-latitude cyclones. The trans-port model is integrated for 17 months for eachsource scenario and the last 12 months are usedfor this study.

2.1. Transport

For this study, NO.. represents the sum of allgaseous and particulate nitrogen compounds,with the exception of N 2O, resulting from bothdirect emission and chemical transformations.This collection of reactive nitrogen compounds istransported as a single tracer that satisfies thefollowing continuity equation:

oRp. ~ V R 0,.at= -Va. 2P. --

fuels. results in an estimated yearly global sourceof 21 x 1012 grams of nitrogen (21 tg N) (Logan,1983). This surface source greatly exceeds theother documented global source, stratosphericinjection. which has a range of 0.5-1.0 tg N (Levyet al., 1980). While a number of other sourceshave been suggested. (lightning. biomass burn-ing. microbial activity) estimates of their globalcontribution are quite uncertain and none arecurrently thought to be as important as fossil fuelcombustion (e.g.. Logan, 1983).

In this study a medium resolution (- 265 kmgrid) general circulation transport model(Mahlman and Moxim. 1978) is used to simulatethe global climatology of reactive nitrogen result-ing from fossil fuel combustion. The parentgeneral circulation model (Manabe et al.. 1974)generates an ensemble of realistic weather pat-terns that compares well with observed yearlyclimatology and with observed meteorology ontime scales ranging from six hour synoptic eventsto seasonal cycles. While regional forecast modelsmay generate more realistic simulations ofindividual synoptic events and provide valuableinsights to specific transport processes anddeposition events (e.g.. Chang et al., 1987), thesemodels have not yet generated climatologiesof concentration and deposition. Such data isneeded for studying both the cumulative impactof acid deposition on a regional environment andthe global nitrogen budget.

First, we simulate the distribution and depo-sition of the global emissions of combustionnitrogen and compare the model's results withavailable observations. Major sources of nitrogenin remote regions are then identified and con-clusions are drawn regarding the impact ofcombustion NO, on the Southern Hemisphere,the possible impact of peroxyacetylnitrate (PAN)on global-scale transport and the potential rolefor other less certain sources of reactive nitrogen.

2. Model description

The GFDL general circulation/transport modelhas 11 terrain following levels with standardheights of 31.4,22.3, 18.8, 15.5, 12.0,8.7,5.5.3.1,1.5. 0.5, 0.08 km and a horizontal grid size ofapproximately 265 km or 2.4°. The off-line trans-port model has a time step of -26 min and is

Tellus 418 (1989), 3

00' UP. R + diffusion

+ source -Welf P. R,urf -<l>p. R. (1)

R is the NOy volume mixing ratio, P. is surfacepressure, V2 is the horizontal wind, 0' = pip. is thevertical coordinate where p is pressure, Welf is thegroup deposition velocity that is 0.0 in all but thelowest level, R,urf is the NOy volume mixing ratioat the surface and <1> is the group precipitationremoval coefficient. The preparation of inputdata, numerical integration techniques and modelcharacteristics have all been described in pre-vious simulations of tropospheric nitrous oxideand ozone and the distribution and deposition ofnuclear debris (Mahlman and Moxim, 1978;Levy et al., 1982; Levy et al., 1985).

"Diffusion" includes both explicitly parameter-ized horizontal and vertical subgrid-scale dif-fusion and an implicit contribution from thecomputational adjustment of negative mixingratios (Mahlman and Moxim, 1978). The stan-dard vertical diffusion term, which dominatesvertical transport in the lower troposphere whenthe model is convectively unstable (dry or moist)on the grid scale, has been described previously(Levy et al., 1982). However, to account forturbulent mixing in the boundary layer when the

258 H. LEVY II AND W. J. MOXIM

(7.9 tg N) are based on a country by countrycompilation of UN fuel use statistics and EPAemission factors (Hameed and Dignon, 1988). Agiven country's emissions are then distributedamong the model grid boxes by population andfurther apportioned among the surface andbottom two levels in the fraction 0.4, 0.4, and 0.2respectively. The distribution of total yearlyemissions at 21.3 tg N is shown in Fig. I.

model is stable on the grid scale. we include anadditional shear-dependent vertical diffusionterm

kv = A L(:)21i'V/i'=I.

where L(=) is 27 m, 17 m and 0.75 m in thebottom three levels respectively and 0.0 above.A = 0.1 gives the best agreement between mid-continental winter profiles for 222Rn simulatedby this transport model and profiles based on ananalysis of available observations (Liu et al..1984).

2.3. ChemistryChemical reactions indirectly control depo-

sition by converting insoluble nitrogen com-pounds to soluble nitric acid and nitrate andapportioning Nay among the various specieswith their differing deposition velocities. Exten-sive gas-phase chemical reaction schemes havebeen developed (e.g., Leone and Seinfeld, 1985),and a simplified version (Stockwell, 1986) hasbeen recently employed in a regional aciddeposition model (Chang et al., 1987). Thoughthese chemical schemes can reproduce smogchamber experiments, they have only recentlybeen compared with the chemistry in thebackground continental troposphere (e.g.,Trainer et al., 1987).

The detailed chemistry is most importantwithin the source region. Nay is transported rela-tively slowly in the boundary layer (100-200km/day) and generally lost to dry deposition in1-2 days, regardless of the chemical partitioning.Once in the free troposphere, a region not directlyaffected by boundary layer processes, removal isless likely and chemical partitioning is again lessimportant. Chemical partitioning may, however,play an important role in the wintertime Arcticwhere PAN is the dominant form of Nay(Bottenheim et al., 1986) and the collective tracermodel underestimates both the effective Naylifetime and its transport.

To realistically simulate the impact of fossilfuel combustion on the global nitrogen budget,the model must generate the correct long-rangetransport of nitrogen combustion products fromthe major source regions. Besides a realisticmeteorology, this requires that the correct frac-tion of combustion nitrogen be available fortransport, after dry and wet removal in the sourceregion. Rather than using a complex chemicalreaction scheme to generate HNO3 formationrates and surface partitionings that reproduce

2.2. SourceThe global source function in eq. (I), "source",

is constructed from a variety of emission inven-tories, fuel use statistics and emission factors, andis held constant in time for each grid box. Thereis no seasonal cycle in the source, and neither theparent GCM, the off-line transport model nor thesource has a diurnal cycle. While Europe, becauseof low demand for cooling in the summer andhigh demand for heating in the winter, is be-lieved to have a seasonal cycle in NO, emissions,it is less than the uncertainty in the absolutesource strength and not important to this study.

The US and Canadian portion of the globalsource is constructed from EPA and Environ-ment Canada emission inventories, and has beendescribed previously (Levy and Moxim, 1987).The area emissions are released at the model'ssurface, and the point sources are apportioned,according to stack height, among the surface andmid-points of the bottom two levels. Of theresulting total yearly emissions (7.5 tg N), 65%are injected at the surface, 20% into the middle ofthe lowest level at a standard height of 80 m, and15% are injected into the middle of the next levelat a standard height of 500 m. While emissions atboth the surface and 80 m enter the same level,the surface emissions experience an immediatereduction due to dry deposition (see Subsection2.4 for more detail).

The NOy emissions for Europe are compiled byEMEP (Eliassen et al., 1988). We release the areasources at the surface and split the point sourcesevenly between the bottom two levels. The result-ing yearly source for Europe is 5.9 tg N with 44%released at the surface and 28% released intoeach of the bottom two levels.

The remainder of the global emissions

Tellus 41B (1989). 3


observed depositions, we transport the reactivenitrogen species collectively as NO, and base themodel's parameterizations for wet and dryremoval on observations of deposition andsurface concentration from the North Americansource region (Levy and Moxim, 1987). Theglobal applicability of this parameterization isdiscussed in Subsection 3.2.

where R,urf(i) is the volume mixing ratio of the ithreactive nitrogen species at the surface and Wi isits measured deposition velocity.

(2) The mixing ratio of the collection of reac-tive nitrogen species at the surface.

By assuming that the bottom half of the lowestlevel is in steady state with surface depositionbalanced by turbulent flux, we can write

R,urf=RII/[I+(weIr/Cdlv.IrI)], (3)

where Cd is the GCM's globally averaged surfacedrag coefficient (0.002), RII is the mixing ratio inthe lowest level and I v.1r I is the model's effectivesurface wind speed. For more details on eq. (3),see Levy et al. (1985). For a recent discussion ofdry deposition in general, see Walcek et al.(1986).

Measurements of Wi over land are available forNOz (0.3-0.4 cm S-I); PAN (0.3-0.4 cm S-I);HNO3 (1.0-2.5 cm S-I) (Cadle et al., 1985;Huebert and Robert, 1985; Wesely et al., 1982;Garland and Penkett, 1976; Walcek et al., 1986;Voldner et al., 1986). Only HNO3 is assumed tobe deposited over snow and water, and Wi isassumed to be 0.0 everywhere for NO. Usingthese values and the seasonal average partition-ing of NO.. based on limited measurements near

2.4. Dry depositionThe loss of gases and particles by their sticking

to or reacting with the earth's surface and itsvegetation as well as their absorption by thatsame vegetation is a continuous process called drydeposition. This process depends on the localmixing ratio of the individual reactive nitrogenspecies at the surface, R,urf(i), and on theirrespective deposition velocities. It'o(i). In themodel, the nitrogen species are transportedcollectively as R(NO..) and dry deposition is aproduct of two quantities:

(1) An effective or group deposition velocitythat depends on the chemical partitioning ofNO..,

(2)Weff=

Tellus 418 (1989), 3


mathematical solution for surface deposition inthe presence of a surface source and produces a7% reduction in the total emissions. Thoughsmall relative to the uncertainties in the absolutesource strength. it does make a significant contri-bution to dry deposition in the source region.

Boulder, CO: Point Arenas, CA; and StateCollege, Pa (Fahey et al., 1986; Fehsenfeld et al.,1988), we calculate weir over snow-free landranging from 0.30 cm S-I in the winter to0.50 cm S-I in the summer. Using data compiledby Logan (1983), we calculate, outside of urbanareas, Weir ranging from 0.6 to 0.25 cm S-I. Whileboth the chemical partitioning and the individualdeposition velocities exhibit considerable vari-ability in both space and time, the seasonalaverage of weir remains bounded over a relativelynarrow range.

We use Weir = 0.3 cm S-I over land and0.1 cm S-1 over sea as constant deposition coef-ficients for the collection of reactive nitrogenspecies comprising NO,. This neglects bothspatial variability due to difference in groundcover and seasonal effects due to changes intemperature, snow cover and chemical partition-ing. Such a simplification is most serious in thewinter high latitudes where even HNO3 is weaklyremoved (Johansson and Granat, 1986).

To determine the model's sensitivity to Weir'separate simulations with the North Americansource were performed for an upper and lowerlimit of Weir over land, 1.0 and 0.1 cm S-I (Levyand Moxim, 1987), and for an upper limit of Weirover ocean, 0.5 cm S-I. Using Weir = 1.0 over land,a value appropriate to HNO3 = NOy, led to avery rapid loss by dry deposition over the sourceregions and resulted in surface concentrationsand wet deposition over export locations thatwere much smaller than observed. The lowerlimit over land, a value appropriate to snowcovered ground with HNO3 no more than0.1 NO" produced surface concentrations in theNorth American source region that were toohigh. To estimate the impact of NO.. aging, Weirover the ocean was increased to 0.5 cm S-I duringthe summer months. Even for the subtropicalNorth Pacific where long-range transport occursin the lower, albeit stable, troposphere just abovethe maritime boundary layer, NO,. was reducedby only 20%. The neglect of NO,. aging shouldnot qualitatively affect our conclusions regardingthe impact of combustion emissions on the nitro-gen budget in the remote maritime atmosphere.

An additional source of dry deposition in themodel is the immediate deposition of a fraction ofthe emissions released at the surface of themodel's lowest level. This is a result of the

2.5. Wet deposition

Precipitation removal. an intermittent processacting throughout the tropospheric column, isalso expressed as a first order loss term in eq. (I).The removal coefficient at level k, <I>t, is given by

<I>t = N Bt[rain/rainG] (4)

where "rain" is the model's accumulated 6-hprecipitation in a given column, "rainG" is theglobal average precipitation rate expressed asdeposition per 6 h, link is a lifetime that, basedon observations of the vertical distribution ofrainfall production, ranges from 20 days at thelowest level to 60 days at the 8.7 km level(Mahlman and Moxim, 1978), and N is a globalempirical parameter. A much more complex for-mulation based on Henry's Law solubility and acloud physics parameterization including rainoutin the cloud and washout below reduces to a formsimilar to eq. (4) for highly soluble gases (privatecommunication, W. L. Chameides). The wetdeposition of radioactive debris from nucleartests in the stratosphere has been successfullysimulated with N = I (Mahlman and Moxim,

1978).For this and a previous study of nitrogen

emissions (Levy and Moxim, 1987), the empiricalparameter is adjusted to N = 9 to bring thesimulated yearly integral of nitrogen depositionin precipitation over the US and Canada intoagreement with the observed deposition of-2 tg N (Logan, 1983; Golomb, 1983; Barrieand Hales, 1984; Galloway and Whelpdale,1987). North America was chosen because it hasthe most complete data and is least affected byoutside sources. We include no seasonal depen-dence in <I>t, in agreement with recent analysis(Summers and Barrie, 1986). While the yearlyintegral has been adjusted, the spatial pattern andseasonal variation has not been fixed. It dependson the model's wind fields and precipitationpatterns and is in good agreement with obser-vations (Levy and Moxim, 1987). While we mightnot expect N = 9 to apply throughout the world,

Tellus 41B (1989), 3

DEPOSITION OF REACrIVE NITROGEN EMITTED BY FOSSIL FUEL COMBUSTION 261

average OH profiles ranging from 1.2 x 10" mol-ecules cm-J at the surface to 0.4 x 106 moleculescm-J at 6 km for the winter and 3.5 x 106 to1.8 X 10" molecules cm-J for the summer (Levy.1974). OH is further reduced by 0.5 to account forthe affect of clouds on 01 D production. and theinitial partitioning of NO, is the same as thatused for the calculations of ~..If. Over the 12-hperiod of the storm. the chemical calculationsfind that. while being lifted. 25 % of the surfaceNO.. is converted to HNOJ and removed in thewinter and 40% in the summer to give an averageof 32.5 %. Over the same period. the precipitationparameterization gives an average NO, removalin the rising parcel of 13% for N= I and 71% forN = 9. This qualitative agreement supports thesimple first-order treatment of wet removal.

there are sufficient deposition data to confirm theapplicability of N = 9 for the Northern Hemi-sphere. This will be discussed in Subsection 3.2on model evaluation.

For a quick qualitative check on our choice ofN, we consider an extremely simplified model ofa mid-latitude precipitation event and calculatethe resulting deposition using both eq. (4) and theoxidation of NO1 to HNO) by OH. Specifyingrain = 3 cm day-I for 12 h. we assume transportonly in the vertical. With a velocity of 10 cm S-I,a volume of air will rise from the surface to4.32 km during the l2-h event.

The HNO) formation during the stormdepends on the fraction of NO} existing as HNO)when the precipitation starts and the rate atwhich NO1 can be converted to HNO) duringthe event. The chemical calculations use daily

3. The global distribution and deposition ofcombulstion nitrogen

After examining the simulated distributionand deposition of global combustion nitrogenemissions, the model's results are compared withthe observed yearly depositions in Europe, near-by export sites and remote locations, and with theyearly concentrations at background sites in theNorth Pacific. Then, the global model is used todetermine the impact of combustion emissions onNOy levels in the Arctic, the North Pacific andthe Southern Hemisphere and to identify thesource regions responsible for each.

3.1. Modelsimulations

Just as was simulated (Levy and Moxim, 1987)and calculated from observation (Galloway andWhelpdale. 1987) for North America, acid rainover the world's source regions accounts for asmall portion (30%) of the global combustionemissions of nitrogen, while dry deposition overthe source regions accounts for 45% and export,mainly over the oceans, for the remaining 25%.

Latitude-height plots of NOy, averaged overlongitude for a month, are shown in Fig. 2 forboth January and July. Previously we observedthat less than 0.1% of the North Americanemissions are transported to the SouthernHemisphere (Levy and Moxim, 1987) and thisappears to hold true for the global source wherethe Southern Hemisphere NO} levels seldom ex.


SOOmb

'a'

- =w'-

, "

-+ ~

i,< 00'

t::= -990mb

,.

WET DEPOSITION

'-/"'\r

EOi

~~~ [?c;J7=:!:5 --'-"

0"~~

Fig, 3, Latitude-longitude plots of the yearly averagesof R(NO..) (ppbv) at 500 mb in the free troposphere and990 mb In the surface layer and of the yearly total wetdeposition of nitrogen (mMole m-2 yr-1),

ceed 0.05 ppbv. Unlike insoluble tracers domi-nated by Northern Hemisphere surface sources(Prather et al., 1987; Plumb and Mahlman,1987), there is no zonally averaged inversion inthe southern hemisphere. Transport through theITCZ rain belt effectively removes most of theNOv before it can be carried up into the middletroPosphere. The strongest seasonal patternoccurs in the northern hemisphere where most ofthe emissions occur. In the winter the 100 pptvcontour spreads to 9O0N though not above500 mb. In the summer, strong convective mixingover the continental source regions lifts thiscontour to 315 mb at mid-latitudes, but there islittle transport north of 75°. The small strato-spheric source of NOy, predicted to support zonala verages of 0.01-0.03 ppbv in the boundary layerand 0.03-0.05 ppbv in the free troposphere (Levyet al., 1980), appears to be more important thanfossil fuel combustion over most of the-SouthernHemisphere and the upper troposphere of theNorthern Hemisphere. Unlike emissions at theground that generally undergo wet removal whenthey are lifted into the free troposphere for long-range transport, NOy from the stratosphere orupper troposphere is transported in dry andrelatively stable air. There is also a possible rolefor nitrogen emissions from biomass burning inthe tropics (Crutzen et al., 1985), although theNOy must survive wet removal processes when itis lifted into the upper troposphere before beingtransported to higher latitudes.

The simulated NOy distributions also havelarge longitudinal gradients, as are shown bylatitude-longitude plots of yearly averaged NOymixing ratio in the boundary layer (990 mb) andthe middle troposphere (500 mb) along with theplot of total yearly wet deposition (see Fig. 3). Asexpected for a short-lived tracer, there is a strongrelationship between the distributions in Fig. 3and the source distribution in Fig. I. In all casesthe maxima occur near the source regions, thoughat 500 mb the gradients are much weaker andNOy is carried down wind. Throughout the tropo-sphere, the NOy mixing ratio is low (0.05 ppbv orless) over the North Pacific and less than0.005 ppbv (see the dotted contour in Fig. 3) overmuch of the southern ocean. The only significantseasonal differences in the fields shown in Fig. 3are a summertime increase in vertical transportdue to increased continental convection, an in-~

creased downwind transport at mid-latitudes inthe winter when the prevailing westerlies arestrongest (particularly over the North Pacific),and increased summertime easterly transport inthe tropics and subtropics.

While there is no inversion in the zonal meanprofiles (Fig. 2), significant regional inversionsdo occur in both hemispheres over the centralAtlantic and Pacific which are down wind of, butdistant from, major source regions (Fig. 4). Mostof the long-range transport in the atmosphere, assimulated in Fig. 4 and inferred from trajectorystudies in the real atmosphere (Merrill et al.,1985), occurs in the free troposphere not in theboundary layer.

Tellus 418 (1989), 3

DEPOSmON OF REACTIVE t~OGEN EMITTED BY FOSSIL FUEL COMBUSTION 263

~~OE"

~

soo

-685.cE-835

~;-,r j j jf6LMillI

180

I ~9AO

c::;r'-.~ "' 0 d'

"

Fig. 5. Map of the European source region with themodel grid superimposed, with the grid boxescontaining the wet deposition sites from Table Iidentified. and with the integrated wet depositionregion outlined.

140W

9901 / I I I I I I1)1 0.1 1.0 10.0

Fig. 4. Local vertical profiles of monthly averagedR(NOj (ppbv) simulated for 3 sites at 48°N latitude

(Japan (140° E). Central Pacific (1801. Eastern Pacific(1400W» in January.

Table Nitrogen wet deposition at European sites

Stationno.

3.2. Model evaluation

While model re:sults have intrinsic value, thekey issue is the model's ability to simulate thebehavior of the real atmosphere and, in this case,real atmospheric transport. To establish this, wecompare the model's simulations with availableobservations.

The wet removal parameter (N = 9) and thecalculated values for Welf' 0.3 cm S-1 over landand 0.1 cm S-1 over ocean, are based on obser-vations in North America. First we check theirglobal applicability. The observed yearly depo-sition over the region of Europe outlined in Fig.5, 1.6 tgjyr (Schaug et al., 1987), compares wellwith the model's self-determined value of1.4 tgjyr. A more detailed comparison is providedin Table I where yearly wet deposition andprecipitation are compared for a number of rep-resentative stations whose locations are noted inFig. 5. The agreement is generally within 25%and seldom worse than a factor of 2. The largestdisagreements between model and observationsoccur in regions with sharp topographical gradi-ents that produce precipitation features not re-solved by the model (i.e., Station 4). As noted bythe reviewer, the model, while frequently simu-lating higher precipitation than observed overEurope, almost always simulates lower wet depo-sition. It appears that the model's parameter-ization underestimates wet deposition in Europeby 15%-20%. While yearly deposition data fromthe third major sourc~ region, Asia, are not

obs. model

123456789

1011

527130

18050

1615738495457

8689447068

1278861679787

22363

46194622

3163013

2923

l

1345163

132515

available, the comparison in Europe is encourag-ing. Our very simple parameterization appears tosimulate realistic deposition in source regionsand, as a result, should provide a realistic effec-tive source for long-range transport.

In Table 2, the model's wet deposition valuesare compared with observations (J. N. Galloway,private communication) from regional stationsno more than 1000-2000 km downwind ofmajor sources (Adrigole, Ireland; Bermuda; BayDeSpoir, Newfoundland; Katherine, Australia;Nova Scotia) and from remote locations only

Tellus 41B (1989), 3

JANUARY

.soN315. ~ '"

';\ ""


Table 2. Nitrogen wet deposition at regional and remote sites

Precip. (cm yr-l)

obs.

Wet dep. (mMole m-2 yr-l)

Station name No. model obs. model

123456789

10

1381492001402947913649

123

1169390

18930-50.9733-99.20

135138

16.09.1

11.26.30.50.54.01.60.21.9

15.010.05.35.40.2-0.7-0.30.1-0.2-0.20.1fI!

Nova ScotiaBay de Espoir, Nfld.Adrigole, IrelandBennudaPoker Flats, AlaskaMauna Loa, HawaiiKatherine, AustraliaCape Point, SATorre de Paine, ChileAmsterdam, Island

.Because of a steep gradient in model precipitation, neighboring boxes were included.

its highly simplified parameterization of wetdeposition.

In Table 3 we compare mixing ratios from themodel's lowest level with observations (Prosperoet al., 1985; Savoie et al., 1988) from a string ofisland stations running from Shemya in theNorth Pacific to Norfolk Island in the SouthPacific (see Fig. 6). The filter measurements arelower limits since water insoluble species, mainlyorganic nitrates and nitrogen oxides are not in-cluded. While some surface measurements of the

affected by long-range transport (AmsterdamIsland; Torre del Paine, Chile; Poker Flat,Alaska; Mauna Loa, Hawaii; Cape Point, SouthAfrica). These sites are located on Fig. 1. Withthe exception of Amsterdam Island, all theremote sites have local topOgraphy that the GCMcannot resolve, and Katherine, Australia, thoughreceiving 91 cm yr-1 of rain, is only a fewhundred kilometers from the desert. As a furtherdifficulty, only a fraction of the yearly depositionis collected and analyzed at the remote sites, andwe find that deposition, at sites stronglyinfluenced by the transport of NOy' is not linearin precipitation.

Model and observation are in excellent agree-ment for 3 of the regional deposition sites. Mostof the disagreement at Adrigole is due to thedifference in local precipitation, not due to errorsin the regional and long-range transport. How-ever, even when the deposition from backgroundnitrogen is subtracted from the Katherine,Australia data and the error in simulatedprecipitation is considered, model and obser-vation are far apart. The local combustion source,see Fig. 1, is weak with no large sources nearby.Unless a major industrial source has been missed,the most likely explanation is a large agriculturalsource which is not part of this study. Theagreement at 4 of the 5 regional sites and the tworemote sites in the Northern Hemisphere is with-in a factor of 2 or better. The model's low valuesmay be due to its lack of explicit transport ofinsoluble end-products such as PAN and/or to

\',.

!

ENEWET ;1

NA~

F~

\ -~

I I I.FAIliNG

IAFU~0.'

~!:" ,- {fl'~A!IT\r

1- / ;

40."60.

Fig. 6. Map showing the approximate location of thePacific network of surface stations measuring totalwater-soluble NO,.

Tellus 418 (1989).3

DEPOSITION OF REACTIVE NITROGEN EMITTED BY FOSSIL FUEL COMBUSTION

Table 3. Surface R<NO:}" (ppbr) in the Pacific

Station Obs. Obs. -Background (0.038 ppbv) Model

ShemyaMidwayOahu. HawaiiEnewetakFanningNauruFunafu..tiSamoaRarotongaNew CaledoniaNorfolk

0.0940.1050.1300.0540.0580.0580.0400.0430.0420.0760.065

0.0560.0670.0920.0160.0200.0200.0020.0050.0040.0380.027

0.0740.0430.033* (O.OSa-o.O6O)t0.0200.0110.0040.0030.0030.004

O.Ola-o.OSOt0.028

.The local Hawaiian source has been eliminated in this simulation.t Because of the presence of a local source, data has been taken from upwind boxes.

3.3. The nitrogen component of arctic hazeArctic haze was first reported in the scientific

literature by Mitchell (1956). Following thedemonstration of its association with mid-latitudecombustion by Rahn et al. (1977), there has beena great deal of research (see Barrie (1986». In thewintertime Arctic, which has a very stableboundary layer, little precipitation, a low tem-perature and a snow or ice covered surface,HNO3 and NO) are deposited slowly (Johanssonand Granat, 1986), and PAN, NO and NO2 allhave very long lifetimes. With the Europeansource region, shown in Fig. I, almost 200 northof its counterpart in North America and withsignificant European emissions inside the ArcticCircle, one might expect the European emissionsto be the major cause of Arctic haze. Modelsimulations (e.g., Barrie et al., 1989) find this tobe true for sulfate, though the relative importanceof emissions north and south of the Arctic Circleis still an issue.

In Fig. 7. the average winter NO} mixingratios in the boundary layer are presented for 4source scenarios: (I) global emissions; (2) NorthAmerican emissions; (3) Asian emissions; (4)emissions north of 600 N. For this seasonal study,the simulations begin in October and areintegrated through March. In general Europeanemissions dominate, though North American andAsian emissions are important in their localArctic regions. Emissions north of 600N con-tribute up to half of the NOy in the EuropeanArctic. Further detail is given in Table 4. We see

insoluble species over the eastern North Pacifichave been as high as 0.03--0.05 ppbv (Singh et al.,1986; Atlas, 1988), NOx in the central Pacific ismuch lower (e.g., McFarland et al., 1979) andPAN is at the measurement limit of -0.005 ppbvover the South Pacific (Singh et al., 1986). There-fore, the insoluble nitrogen species are not ex-pected to make a major contribution to thesurface, measurements of NO} over the CentralPacific. As was the case for deposition in mari-time locations, the surface measurements arecorrected for the background soluble nitrogen(0.038 ppbv) measured at extremely remote mari-time sites (Savoie et al., 1988).

With the exception of Nauru, model and obser-vation agree within a factor of 2. A more seriousdifference is the latitude gradient in the NorthernHemisphere. While the model lacks explicittransport of PAN and a time-dependent Welf' theShemya observations are uncertain because of alimited record (Prospero et al., 1985) and those atOahu, though only taken from on-shore winds,are made in a region of relatively high emissions.At this time it is not clear which latitude-gradient is correct. However, measured surfaceconcentrations of Asian dust increase with lati-tude (Uematsu et al., 1985), just as we simulatedfor Asian emissions of NO.,.

Although there are clearly quantitative errorsin the model's simulation of long-range transportand deposition. the degree of agreement, giventhe lack of explicit chemistry and the simpleparameterizations of deposition, is reassuring.

Tellus 41B (1989), 3


NORTH AMERICAN SOURCEGLOBAL SOURCE

,:","-.

~y~~;

, :,,".

180 ~~~E~~) [)

'"'):/ ~.

-:,SV'-')" (].I

'\ ,- '.~ .''(j f\~ ,,-,'',~s ,

ASIAN SOURCE SOURCE NORTH OF 60oN

,'<"'"~~:~ ~ : ."'...

...::::

/

I'

-1'I,:.,-

11::: :.1':':

: : : :,: : : :

;:~:~f180

,-

,-p:;~~~.;

;.::;:-:~1~ 1;~~~t~!:J:-

:.Q~~::...,

180..::",::;;.; ~ ~{

:~..;:. ::::~:...:::.02:.~.\~.~"

:K

,~ .., :-~:\:...'.;,

::~...';;'-'; ::f ,::f~

I':~~~-_.f:!~:..:_.:~;"";' -~';~ii~:,#jC:"f~f~~:_:;;'"Fig. 7. Polar stereographic plots of the winter (Jan.-Feb.-Mar.) average R(NO) (ppbv) at 940 mb (standardheight of 500 m) for 4 sources: global emissions; North American emissions; Asian emissions; global C"~;..;nn.north of 600 N.

that emissions inside the Arctic are responsiblefor half of the total NOy at the surface, while bothEuropean and North American emissions playimportant roles at higher levels. Transport tothese levels in the Arctic is the result of winterextratropical cyclones that generally form overthe northern US and Canada, the North Atlanticand northern Europe. The general pattern shownin Table 4 for the NOy component of Arctic hazeis in harmony with previous simulations forsulfur (e.g., Barrie et al., 1989).

The transport model predicts a strong longi-tudinal gradient in the NOy with mixing ratioslower than 0.1 ppbv in northern Alaska andgreater than I ppbv in northern Scandinavia andthe Soviet Union (see Fig. 7). This pattern isqualitatively similar to the one reported by Barrie(1986) for sulfate. Whether the weaker sulfategradient is due to quantitative differences in theemission patterns, differences in the wintertimeoxidation rates, defects in the model's transportor the larger weft" for NOy is not clear. The slower

Tellus 41B (1989), 3

DEPOSITION OF REACTIVE NITROGEN EMITTED BY FOSSIL FUEL COMBUSTION267

Table 4. The o,~ contribution to NO, in the Arctic by source region

oxidation rate expected for SO~ would allow morezonal mixing of the sulfur species, while NO}.'shigher Welf decreases its atmospheric lifetime andunderstates the degree of zonal mixing.

The measurements of NO} in the Arctic arequite limited. Bottenheim and Gallent (1986)report 0.25~.35 ppbv NO, at Alert in late winter,most of which is PAN and is expected to have along atmospheric lifetime in the winter Arctic,while the model only gives 0.15~.20 ppbv ofNO}. When Welf was set to 0.0 in the Arctic, thesimulated levels of NOy exceeded the obser-vations of Bottenheim and Gallent (1986). Themodel's apparent overestimation of dry depo-sition in the winter Arctic and its lack of explicittransport for PAN exaggerates the gradientbetween North America and Europe and under-estimates the total amount present. However, thequalitative pattern of distribution and the domi-nance of the European sources is not affected.Currently, a multiple-species model, with explicittransport of PAN and a variable wolf' is beingdeveloped to study the nitrogen component ofArctic haze in detail.

00 .~ .0 e e ~ ~ ~ i oj:J0

~

00

-e'..-'ic5'e~~e~I I I I I I I I I I I I

JFMAMJJASNOD

j0 Model

.ObservatHJn

Fig. 8. Comparisons of simulated and oberved seasonalvariations of surface NOy at 5 stations in the NorthPacific as given by plots of 'R'(No:Jmn/~. Theopen circles are the model's values and the filled circlesare observations.

3.4. NO, OL'er the North PacificThe simulated yearly average NOy levels at the

5 surface sites in the North Pacific (see Fig. 6)have already been compared with observations inTable 3 and discussed in Subsection 3.2. Thesimulated and observed seasonal cycles are com-pared in Fig. 8 by plotting the ratio of themonthly mean to the yearly mean. The observedseasonal cycles are based on 2-7 years of obser-vations (Savoie et al., 1988). While the maximaand minima may be shifted by a month, the basicpatterns are well simulated. In general, the modeldoes show a larger winter/spring maximum thanis observed.

By turning off separate source regions we find

Tellus 418 (1989), 3


AUGUST

1"..-""--'""' .'

, 0,' ~20

.'"'"

)~

",s~i..~~I-:" "

"0'

Fig. 9. Latitude-longitude plot of the average Augustmixing ratio for NO.. in the surface layer over thePacific Ocean. R_mb(NO..>mn (ppbv). as simulated bythe transpon model with a global source (F identifiesFanning and M identifies Midway).

that Asian emissions supply two-thirds of theNO.. over the North Pacific. The winter/springmaximum has also been observed at all 5 sites fordust transported from Asia (Uematsu et al.,1983). The second maxima at Oahu and Fanningare due to summertime transport from sources inthe US. In the simulation, the summertime trans-port to Oahu is somewhat greater than observed,while the tongue of maximum NOy (see Fig. 9)remains just a few hundred km north of Fanning.A southward shift of one grid box, well withininterannual variability, would bring the simu-lation much closer to observed. High levels ofNO) have also been measured at Mauna Loa,Hawaii, in air free from contamination by eitherlocal emissions or the maritime boundary layer(Galasyn et al., 1987), and model simulationsidentify both springtime transport from Asia andsummertime transport from the US.

3.5. Combustion NO.. in the southern hemisphere

Figs. 2 and 3 in Subsection 3.l both show largeinterhemispheric gradients and imply that thereis little interhemispheric transport of NO..,.Examining the remote Southern Hemisphere sitesin Tables 2, 3, we see that the model can explainat most 10°;{; of the observed background depo-sitions (-2 mMole m-! yr-l) and surface concen-trations (- 0.04 ppbv) at maritime sites. Themodel does account for the excess NO.. at NewCaledonia and Norfolk Island resulting fromtransported Australian emissions, though itcannot account for the observed deposition atKatherine, Australia. Cape Point, South Africa isessentially a maritime site and Torr de Paine,

Chile is an inland site in the rain shadow of theAndes. If, based on the model's performance inthe Northern Hemisphere, we accept that itslong-range transport in the Southern Hemisphereis relatively realistic. we are left with threepossible explanations for the background nitro-gen: (I) explicit long-range transport of insolubleend-products such as PAN; (2) alternative localand regional sources of NO,; (3) downward trans-port from the lower stratosphere and uppertroposphere.

The current model does not explicitly allow forthe long-range transport of an insoluble end-product such as PAN. Current observations findvery little PAN in the maritime boundary layer ofeither hemisphere but as much as 0.1 ppbv overthe North Pacific in the free troposphere. Theonly free troposphere measurements in the South-ern Hemisphere are from a few flights over theSouth Atlantic between South America andAfrica that find very little PAN (less than0.01 ppbv) outside of local pollution (Rudolph etal., 1987). While a chemical scheme that allowsfor the formation of PAN should increase NO)lev~ls over the North Pacific, it is not clear, givena lifetime of 2 h at 20°C, 2 days at O°C and 14days at -10°C, that PAN will survive transportthrough the ITCZ and on to the South Pacific.However, we are developing a model with anexplicit chemistry and three transported tracers(nitrogen oxides, HNO3 and NO), PAN) toexamine this possibility.

Unless there is a means to concentrate organicnitrogen in the surface layer of seawater, seasaltaerosol would not appear to be a major source.However, Prospero et al. (1985) and Savoie et al.(1988) have suggested lightning, long-range trans-port of emissions from biomass burning andstratospheric injection as possibilities. Lightningis very difficult to quantify and much morevariable in space and time than the backgroundNO... The long-range transport of nitrogenemissions from biomass burning in the tropics,most likely in the form of PAN, is a possibility,though there are no observations to support thistransport.

Past calculations of downward transport fromthe stratosphere (Levy et al., 1980) explainapproximately half of the background. Aninteresting point is the qualitative similaritybetween seasonal cycles at Samoa for surface

Tellus 41B (1989), 3


measurements of 0) (Oltmans, 1981), which isknown to be supplied from the stratosphere (Levyet al., 1985) and surface measurements of NO.(Savoie et al., 1988). Measurements of NO. andPAN in the free troposphere over the SouthPacific would help identify transport from aloft,be it long-range transport of PAN from thenorthern mid-latitudes and tropical regions ofbiomass burning, or NO, transported from thestratosphere and the upper troposphere.

4. Conclusions

combustion in the Northern Hemisphere is trans-ported into the Southern Hemisphere. The littlethat is, as well as the prescribed SouthernHemisphere emissions, can explain no ~ore than10% of the background nitrogen that dominatesboth the nitrogen budget of the SouthernHemisphere and the NOy climatology outside ofcontinental source regions. While the long-rangetransport of PAN, local marine biogenic pro-duction, and lightning sources are all possibleexplanations, we predict that stratospheric injec-tion is a major source of background NO, in theSouthern Hemisphere.

Measurements of individual reactive nitrogenspecies and total NO, are needed throughout thefree troposphere, particularly in the SouthernHemisphere where there are almost none.

Future model developments will include chem-istry. transport of individual nitrogen com-pounds, PAN in particular, and a seasonal andsurface dependent dry deposition parameter-ization.

5. Acknowledgements

We wish to thank Osten Hov, Sultan Hameedand Jane Dignon for providing emission data,Charles W. Black and Louisa A. Bogar for con-structing the global emission field, Dennis Savoieand James Galloway for providing unpublishedobservations and Anton Eliassen, DouglasWhelpdale, Dennis Savoie and James Gallowayfor many helpful comments and discussion.

Using a medium-resolution transport modelwith realistic meteorology, no explicit chemistry,and simple parameterized wet and dry depo-sition, we have simulated the contribution ofNO, emissions from fossil fuel combustion to theglobal NOy climatology and we conclude thefollowing.

(1) Asian emissions, which produce a spring-time maximum from Shemya to Fanning, are themajor source of NOy in the North Pacific. Inaddition, there is summertime transport of NO,from the US out into the central subtropicalPacific.

(2) European emissions are the dominantsource of the nitrogen component of Arctic haze.A large longitudinal gradient (factor of 10 ormore) is predicted with the minimum overAlaska and the maximum over northern Scandi-navia and the Soviet Union.

(3) Almost none of the NO, from fossil fuel

REFERENCES

Atlas, E.1988. Evidence for )C] alkyl nitrates in ruraland remote atmospheres. Nature 331,426-438.

Barrie, L. A. 1986. Arctic air pollution: an overview ofcurrent knowledge. Atmos. Environ. 20, 643-663.

Barrie, L. A. and Hales, J. M. 1984. The spatialdistributions of precipitation acidity and major ionwet deposition in North America during 1980. Tellus36B,333-355.

Barrie, L. A., Olson, M. P. and Oikawa, K. K. 1989.The flux of anthropogenic sulfur into the Arcticfrom mid-latitudes in 1979/80. Atmos. Environ., inpress.

Bottenheim, J. W., Gallant, A. G. and Brice, K. A.1986. Measurements of NO.. species and 0] at 82°Nlatitude. Geophys. Res. Lett.I3, 113-116.

Cadle, S. H., Dasch, J. M. and Mulawa, P. A. 1985.Atmospheric concentrations and the depositionvelocity to snow of nitric acid, sulfur dioxide andvarious particulate species. A/mos. Environ. 19, 1819-1827.

Chang, J. S., Brost, R. A., Isaksen, I. S. A., Middleton,P., Stockwell, W. R. and Walcek, C. J. 1987.RADM, a three-dimensional Eulerian acid depo-sition model. Part I: Physical concepts and modelformulation. J. Geophys. Res. 92, 14681-14700.

Crutzen, P. J. 1974. Photochemical reaction initiatedby and influencing ozone in unpolluted troposphericair. Tel/us 26,58-70.

Crutzen P. J., Delany, A. C. Greenberg, F.,Haagenson, p, Heidt, L., Jueb, R., Pollock, W.,

Tellus 41B (1989), 3


Seiler, W., Wai"tburg, A. and Zimmerman, P. 1985.Tropospheric chemical composition measurements inBrazil during the dry season. J. Atmos. Chem. 2, 233-256.

Eliassen, A., Hov, 0., Iversen, T., Saltbones, J. andSimpson, D. 1988. Estimates of airborne transboundarytransport of sulphur and nitrogen over Europe. Oslo,Norway: The Norwegian Meteorological InstituteMeteorological Synthesizing Centre, West, of EMEP.

Fahey, D. W, Hubler, G., Parrish, D. D., Williams,E. J., Norton, R. B. and Ridley, B. A. 1986. Reactivenitrogen species in the troposphere. J. Geophys. Res.91,9781-9793.

Fehsenfeld, F. C., Parrish, D. D. and Fahey, D. W.1988. The measurement of NO, in the non-urbantroposphere. In Tropospheric ozone (ed. I. S. A.Isaksen). Dordrecht: D. Reidel Publishing Co., 185-215.

Galasyn, J. F., Tschudy, K. L. and Huebert, B. J. 1987.Seasonal and diurnal variability of nitric acid vaporand ionic aerosol species in the remote free tropo-sphere at Mauna Loa, Hawaii. J. Geophys. Res. 92,3105-3113.

Galloway, J. N. and Whelpdale, D. M. 1987. WATOX-86 overview and western North Atlantic Ocean SandN atmospheric budgets. Globol Biogeochem. Cycles 1,261-281. .

Garland, J. A. and Penkett, S. A. 1976. Absorption ofperoxy acetyl nitrate and ozone by natural surfaces.Atmos. Environ. 1O, 1127-1131.

Golomb, D. 1983. Acid deposition-precursor emissionrelationship in the northeastern USA. The effective-ness of regional emission reduction. Atmos. Environ.17, 1387-1390.

Hameed, S. and Dignon, J. 1988. Changes in thegeographical distributions of global emissions of NO,and SO.. from fossil-fuel combustion between 1966and 1980. Atmos. Environ. 22, 441-449.

Huebert, B. J. and Robert, C. H. 1985. The drydeposition of nitric acid to grass. J. Geophys. Res. 90,2085-2090.

Johansson, C. and Granat, L. 1986. AN experimentalstudy of the dry deposition of gaseous nitric acid tosnow. Atmos. Environ. 20,1165-1170.

Leighton, P. A. 1961. Photochemistry of Air Pollution.New York: Academic Press, 300 pp.

Leone, J. A. and Seinfeld, J. H. 1985. Comparativeanalysis of chemical reaction mechanisms forphotochemical smog. Atmos. Environ. 19, 437-464.

Levy II, H. 1971. Normal atmosphere: large radicaland formaldehyde concentrations predicted. Science173, 141-143.

Levy II, H., Mahlman, J. D. and Moxim, W. J. 1982.Tropospheric N:O variability. J. Geophys. Res. 87,3061-3080.

Levy II, H., Mahlman, J. D. and Moxim, W. J. 1980. Astratospheric source of reactive nitrogen in theunpolluted troposphere. Geophys. Res. Lett. 7, 441-444.

Levy II, H., Mahlman, J. D., Moxim, W. J. and Liu,S. C. 1985. Tropospheric ozone: the role of transport.J. Geophys. Res. 90, 3753-3772.

Levy II, H. and Moxim, W. J. 1987. The fate of US andCanadian combustion nitrogen emissions. Nature328,414-416.

Levy II, H. 1974. Photochemistry of the troposphere. InAdvances in Photochemistry. Vol. 9 (ed. J. N. Pitts, Jr.,G. S. Hammond and K. Gollnick). New York: JohnWiley and Sons, Inc., 369-524.

Liu, S. C., McAfee, J. R. and Cicerone, R. J. 1984.Radon 222 and tropospheric vertical transport. J.Geophys. Res. 89, 7291-7297.

Logan, J. 1983. Nitrogen oxides in the troposphere:global and regional budgets. J. Geophys. Res. 88,10785-10807. .

Mahlman, J. D. and Moxim, W. J. 1978. Tracersimulation using a global general circulation model:results from a midlatitude instantaneous sourceexperiment. J. Atmos. Sci. 35, 134(}-1374.

Manabe, S., Hahn, D. G. and Holloway, Jr., J. L. 1974.The seasonal variation of the tropical circulation assimulated by a global model of the atmosphere. J.Atmos. Sci. 31, 43-83.

Manabe, S. and Holloway, Jr., J. L. 1975. The seasonalvariation of the hydrologic cycle as simulated by aglobal model of the atmosphere. J. Geophys. Res. 80,1617-1649.

McFarland, M., Kley, D., Drummond, J. W.,Schmeltekopf, A. L. and Winkler, R. H. 1979. Nitricoxide measurements in the equatorial Pacific region.Geophys. Res. Lett. 6, 605~8.

Merrill, J. T., Bleck, R. and Avila, L. 1985. Modelingatmospheric transport to the Marshall Islands. J.Geophys. Res. 90, 12927-12936.

Mitchell, M. 1956. Visual range in the polar regionswith particular reference to the Alaskan Arctic. J.Atmos. Terr. Phys. Special Supplement, 195-211.

National Res. Council. 1983. Acid deposition: atmos-pheric processes in Eastern North America. a Review ofcurrent scientific understanding. Washingotn, D.C.:National Academy Press, 375 pp.

Oltmans, S. J. 1981. Surface ozone measurements inclean air. J. Geophys. Res. 86, 1174-1180.

Plumb, R. A. and Mahlman, J. D. 1987. The zonallyaveraged transport characteristics of the GFDLgeneral circulation/transport model. J. Atmos. Sci. 44,298-327.

Prather, M. J., McElroy, M., Wofsy, S., Russell, G. andRind, D. 1987. Chemistry of the global troposphere:fluorocarbons as tracers of air motion. J. Geophys.Res. 92, 6579-6613.

Prospero, J. M., Savoie, D. L., Nees, R. T., Duce, R. A.and Merrill, J. 1985. Particulate sulfate and nitrate inthe boundary layer over the North Pacific Ocean. J.Geophys. Res. 90, 10586-10596.

Rahn, K. A., Borys, R. D. and Shaw, G. E. 1977. TheAsian source of Arctic haze bands. Nature 268, 713-715.

Tellus 418 (1989), 3

271DEPOSITION OF REACTIVE NITROGEN EMITTED BY FOSSIL FUEL COMBUSTION

Trainer. M.. Williams. E. J.. Parrish. D. D.. Buhr.M. P.. Allwine. E. J.. Westberg. H. H.. Fehsenfeld.F. C. and Liu. S. C. 1987. Models and observationsof the impact of natural hydrocarbons on ruralozone. Nature 329. 705-707.

Uematsu. M.. Duce. R. A.. Prospero. J. M.. Chen.L. Q.. Merrill. J. T. and McDonald. R. L. 1983. Thetransport of mineral aerosol from Asia over theNorth Pacific Ocean. J. Geophys. Res. 88. 5343-5352.

Voldner. E. C.. Barrie. L. A. and Sirois. A. 1986. Aliterature review of dry deposition of oxides of sulfurand nitrogen with emphasis on long-range transportmooelling in North America. Atmos. Environ. 20.2101-2123.

Walcek. C. J.. Brost. R. A.. Chang, J. S. and Wesely,M. L. 1986. SO~. sulfate. and HNO) depositionvelocities computed using regional land use andmeteorological data. Atmos. Environ. 20. 949-964.

Wesely M. L.. Eastman. J. A.. Stedman. D. H. andYalvac. E. D. 1982. An eddy correlation measure-ment of NO~. Atmos. Environ. /6.815-820.

Rudolph. J.. Vierkorn-Rudolph. B. and Meixner. F. X.1987. Large-scale distribution of peroxyacetylnitrateresults from the STRATOZ III flights. J. Geophys.Res. 92.6653-6661.

Savoie. D. L., Prospero, J. M. and Graustein, W. C.1988. Background concentrations of nitrate in themarine boundary layer (Abstract). Eos Trans. AGU69.317.

Schaug, J., Hansen, J. E.. Nodop, K.. attar. B. andPacyna. J. M. 1987. Summar., Report from the Chemi-cal Co-ordinating Centre for the Third Phase of EM EP.Lillestrom, Norway: Norwegian Institute for AirResearch.

Singh, H. B., Salas, L. J. and Viezee, W. 1986. Theglobal distribution of peroxyacetyl nitrate. Nature321,588-591.

Stockwell, W. R. 1986. A homogeneous gas phasemechanism for use in a regional acid depositionmodel. Atmos. Environ. 20, 1615-1632.

Summers, P. W. and Barrie, L. A. 1986. The spatial andtemporal variation of the sulfate to nitrate ratio ineastern North America. Water. Air. and Soil Pollution30, 275-283.

Tellus 418 (1989), 3

simulated global distribution and deposition of reactive ... · tel/us (1989),4/8,256-271 simulated...

Documents