Simulating emission and chemical evolution of coarse sea …...J. T. Kelly et al.: Simulating...

Geosci Model Dev 3 257ndash273 2010wwwgeosci-model-devnet32572010copy Author(s) 2010 This work is distributed underthe Creative Commons Attribution 30 License

GeoscientificModel Development

Simulating emission and chemical evolution of coarse sea-saltparticles in the Community Multiscale Air Quality (CMAQ) model

J T Kelly1 P V Bhave1 C G Nolte1 U Shankar2 and K M Foley1

1Atmospheric Modeling and Analysis Division National Exposure Research Laboratory Office of Research andDevelopment US Environmental Protection Agency RTP NC USA2Institute for the Environment University of North Carolina Chapel Hill NC USA now at Planning and Technical Support Division Air Resources Board California Environmental Protection AgencySacramento CA USA

Received 18 November 2009 ndash Published in Geosci Model Dev Discuss 4 December 2009Revised 23 March 2010 ndash Accepted 25 March 2010 ndash Published 8 April 2010

Abstract Chemical processing of sea-salt particles incoastal environments significantly impacts concentrations ofparticle components and gas-phase species and has implica-tions for human exposure to particulate matter and nitrogendeposition to sensitive ecosystems Emission of sea-salt par-ticles from the coastal surf zone is known to be elevated com-pared to that from the open ocean Despite the importance ofsea-salt emissions and chemical processing the US EPArsquosCommunity Multiscale Air Quality (CMAQ) model has tra-ditionally treated coarse sea-salt particles as chemically in-ert and has not accounted for enhanced surf-zone emissionsIn this article updates to CMAQ are described that enhancesea-salt emissions from the coastal surf zone and allow dy-namic transfer of HNO3 H2SO4 HCl and NH3 betweencoarse particles and the gas phase Predictions of updatedCMAQ models and the previous release version CMAQv46are evaluated using observations from three coastal sites dur-ing the Bay Regional Atmospheric Chemistry Experiment(BRACE) in Tampa FL in May 2002 Model updates im-prove predictions of NOminus3 SO2minus

4 NH+

4 Na+ and Clminus con-centrations at these sites with only a 8 increase in run timeIn particular the chemically interactive coarse particle modedramatically improves predictions of nitrate concentrationand size distributions as well as the fraction of total nitrate inthe particle phase Also the surf-zone emission parameteri-zation improves predictions of total sodium and chloride con-centration Results of a separate study indicate that the modelupdates reduce the mean absolute error of nitrate predictions

Correspondence toJ T Kelly(jkellyarbcagov)

at coastal CASTNET and SEARCH sites in the eastern USAlthough the new model features improve performance rela-tive to CMAQv46 some persistent differences exist betweenobservations and predictions Modeled sodium concentrationis biased low and causes under-prediction of coarse particlenitrate Also CMAQ over-predicts geometric mean diameterand standard deviation of particle modes at the BRACE sitesThese over-predictions may cause too rapid particle dry de-position and partially explain the low bias in sodium predic-tions Despite these shortcomings the updates to CMAQ en-able more realistic simulations of chemical processes in envi-ronments where marine air mixes with urban pollution Themodel updates described in this article are included in thepublic release of CMAQv47 (httpwwwcmaq-modelorg)

1 Introduction

Sea-salt particles emitted by oceans contribute significantlyto the global aerosol burden on a mass basis (Seinfeld andPandis 1998 Lewis and Schwartz 2004) Sea-salt emis-sions are also important on a number basis and impact con-centrations of cloud condensation nuclei (Pierce and Adams2006) Upon emission sea-salt particles have chemical com-position similar to their oceanic source (eg major ionsNa+ Mg2+ Ca2+ K+ Clminus SO2minus

4 Tang et al 1997)but they are processed chemically during atmospheric trans-port For instance a number of studies have reported up-take of gaseous acids by sea salt eg nitric acid (Gard etal 1998 and references therein) sulfuric acid (McInnes etal 1994) dicarboxylic acids (Sullivan and Prather 2007)and methylsulfonic acid (Hopkins et al 2008) Given the

Published by Copernicus Publications on behalf of the European Geosciences Union

258 J T Kelly et al Simulating emission and chemical evolution of coarse sea-salt particles

large contribution of sea salt to atmospheric particulate mat-ter (PM) the emission and chemical evolution of sea-salt par-ticles must be represented accurately by models

The diameter of sea-salt particles spans several orders ofmagnitude but the peak in the mass distribution is usu-ally in the coarse size range (aerodynamic diameterDaerogt25 microm) (eg Keene et al 2007) Uptake of gaseousspecies by coarse sea-salt particles reduces their availabilityfor condensation on fine particles and can potentially reducethe mass concentration of PM25 (PM with Daerole 25 microm)Uptake by coarse sea salt can also significantly reduce theconcentration of nitric acid in environments where the for-mation of particulate ammonium nitrate is unfavorable (egammonia-limited or high-temperature) Associations be-tween coarse particle nitrate and sea salt have been observedin both coastal (eg Hsu et al 2007) and rural (eg Lee etal 2008) areas

Sea-salt emissions are enhanced in the coastal surf zonecompared to the open ocean and result in elevated concentra-tions near the coast (de Leeuw et al 2000) During advec-tion toward land sea salt is often exposed to anthropogenicemissions from shipping lanes (Osthoff et al 2008 Simonet al 2009) and coastal urban centers (Nolte et al 2008)Considering that many coastal areas are densely populated(Nicholls and Small 2002) chemical modification of sea-salt particles by acidic gases could result in significant humanexposure to anthropogenic PM10 (PM with Daerole 10 microm) incoastal environments This exposure is a concern in light ofassociations between increases in coarse particle concentra-tions and adverse health effects (Brunekreef and Forsberg2005 Sandstrom et al 2005 Volckens et al 2009)

Despite the significance of sea-salt emissions and chemi-cal transformations some prominent air quality models havenot treated sea-salt particles (eg Bessagnet et al 2004Grell et al 2005) Other models have included emissionsof sea-salt particles but have not simulated their chemi-cal interactions with gas-phase species (eg Foltescu et al2005 Smyth et al 2009) The US EPArsquos Community Mul-tiscale Air Quality (CMAQ) model has included online cal-culation of sea-salt emissions from the open ocean since ver-sion 45 but has not accounted for enhanced emissions fromthe coastal surf zone and has treated coarse sea-salt particlesas dry and chemically inert (Sarwar and Bhave 2007)

Studies that have simulated the chemical evolution of sea-salt particles have used alternative models to CMAQ (eg Ja-cobson 1997 Lurmann et al 1997 Meng et al 1998 Sunand Wexler 1998b Sartelet et al 2007 Athanasopoulou etal 2008 Pryor et al 2008) or variants of CMAQ such asCMAQ-MADRID (Zhang et al 2004) These studies of-ten suffered from simple estimates of sea-salt emissions ordid not evaluate model results against measurements of size-segregated PM composition (ie size-composition distribu-tions) Spyridaki et al (2006) did evaluate size-compositiondistributions but did not account for enhanced emissionsof sea salt from the coastal surf zone Kleeman and Cass

(2001) modeled surf-zone emissions but only evaluated size-composition distributions for particles withDaerolt 18 micromA recent example of a CMAQ variant that treats chemicalprocessing of sea salt is CMAQ-UCD (Zhang and Wexler2008) This model was developed for application in theBay Regional Atmospheric Chemistry Experiment (BRACE)(Nolte et al 2008) Although CMAQ-UCD performed wellin that study the model is not suitable for many applicationsbecause its run speed is about 8ndash10 times slower than thestandard version of CMAQ used for regulatory applicationsDespite the numerous modeling efforts described above aneed exists for a computationally-efficient treatment of sea-salt emissions and chemical evolution in a model where re-sults capture the size-composition distributions observed incoastal environments

The BRACE study was conducted to (1) improve estimatesof atmospheric nitrogen deposition to Tampa Bay FL (2) ap-portion nitrogen deposition to local and non-local sourcesand (3) assess the impact of utility controls on nitrogen de-position to Tampa Bay (Atkeson et al 2007) Excessive ni-trogen addition to waterways from the atmosphere and landcan produce eutrophic conditions detrimental for aquatic life(eg low dissolved O2 and high opacity) In 2004 65 ofassessed systems in the continental US had moderate to higheutrophic conditions (Bricker et al 2007) Due to the differ-ent deposition velocities of gases and particles condensationof HNO3 and NH3 on coarse sea salt can alter nitrogen de-position to sensitive ecosystems (Pryor and Sorensen 2000Evans et al 2004) Studies that apportion nitrogen depo-sition to potentially controllable sources could benefit frommodels that accurately and efficiently calculate the chemicalprocessing and deposition of sea salt

Air quality models require good predictions of particlesize distributions to accurately predict dry deposition Ac-curate size distributions are also important to the ongoingdevelopment of an inline photolysis module for CMAQ (Fo-ley et al 2010) and the coupled meteorology and chemistrymodel WRF-CMAQ (Pleim et al 2008) which calculatethe impact of atmospheric particles on radiative transfer andclouds Lung dosimetry models also require information onparticle size because deposition patterns in the lung dependstrongly on particle diameter in addition to flow variables andlung morphometry (Asgharian et al 2001) Due to the reg-ulatory emphasis on mass-based PM concentrations particlesize distributions from the CMAQ model are rarely evaluatedagainst observations In cases where they have been evalu-ated (Elleman and Covert 2009 Park et al 2006 Zhang etal 2006) the focus has been on number or volume distribu-tions of fine particles The availability of size-resolved PMcomposition measurements from the BRACE campaign thatspan two-orders of magnitude (018lt Daerolt 18 microm) pro-vides an opportunity to evaluate CMAQ predictions of size-composition distributions in a coastal urban environment

In this study CMAQ is updated for the version 47 publicrelease to include enhanced emissions of sea-salt particles

Geosci Model Dev 3 257ndash273 2010 wwwgeosci-model-devnet32572010

J T Kelly et al Simulating emission and chemical evolution of coarse sea-salt particles 259

from the coastal surf zone and a chemically interactivecoarse particle mode that enables dynamic transfer of HNO3H2SO4 HCl and NH3 between coarse particles and the gasphase The updated version of CMAQ is applied to theTampa Bay region and predictions of size-composition dis-tributions and gas-particle partitioning are evaluated againstmeasurements from the BRACE campaign in May 2002Results from the updated model are compared with resultsfrom CMAQv46 to demonstrate the model improvementsand computational efficiency Comparisons with observa-tions are used to identify areas for future model development

2 Modeling

21 Aerosol modeling

A brief description of CMAQrsquos aerosol module is given heresee Binkowski and Roselle (2003) for further details CMAQrepresents the atmospheric particle distribution as the super-position of three log-normal modes The ISORROPIAv17thermodynamic model (Nenes et al 1998) is used to equili-brate inorganic components of the two fine modes with theirgaseous counterparts In CMAQv46 and prior model ver-sions the coarse particle mode is treated as dry and chemi-cally inert with a fixed geometric standard deviation (GSD)of 22 These assumptions have been relaxed in the updatesfor CMAQv47 described in this paper In the remainder ofSect 21 the dynamically interactive coarse particle modeused in CMAQv47 is described along with changes to thetreatment of particle-distribution GSDs The parameteriza-tion of sea-salt emissions from the coastal surf zone used inCMAQv47 is described in Sect 22 Additional scientificupdates to CMAQ that were released in version 47 are de-scribed by Foley et al (2010)

211 Dynamically interactive coarse particle mode

Wexler and Seinfeld (1990) demonstrated that time scalesfor gas-particle equilibration are long compared to those ofother processes for certain atmospheric conditions Allen etal (1989) and Wexler and Seinfeld (1992) found evidence ofdepartures from equilibrium possibly due to mass-transferlimitations in field studies of gas and particle systems Mengand Seinfeld (1996) calculated that submicron particles inthe atmosphere rapidly attain equilibrium with the gas phasebut that coarse particles generally exist in non-equilibriumtransition states Evidence from these and other studies sug-gests that models of coarse sea-salt chemistry must simulategas-particle mass transfer rather than assuming instantaneousgas-particle equilibrium

Simulating the dynamics of gas-particle mass transfer ischallenging because some components of the system equi-librate significantly faster than others and require small in-tegration steps to be used for the entire system (ie the

condensation-evaporation equations are stiff) Since compo-nent vapor pressures must be determined at each step usinga computationally-intensive thermodynamic module smalltime steps make the integration impractical for many air qual-ity applications A number of studies have proposed approx-imate techniques for expediting this integration eg Sunand Wexler (1998a) Capaldo et al (2000) Jacobson (2005)Zhang and Wexler (2006) and Zaveri et al (2008) Theldquohybrid approachrdquo of Capaldo et al (2000) and Pilinis etal (2000) is adopted in CMAQv47 since it has been usedwith success in a number of previous studies (eg Gaydos etal 2003 Koo et al 2003 Zhang et al 2004 Sartelet et al2006 2007 Athanasopoulou et al 2008)

Two main sources of stiffness must be overcome whenintegrating the condensation-evaporation equations Firstfine particles equilibrate relatively quickly with the gas phasecompared to coarse particles due in part to the higher surfacearea-to-volume ratios of fine particles Second the hydrogenion concentration changes faster than concentrations of othercomponents because the flux of hydrogen ion is determinedby the sum of the fluxes of H2SO4 HNO3 HCl and NH3and the hydrogen ion concentration is relatively small (Sunand Wexler 1998a Zaveri et al 2008) To minimize stiff-ness two key assumptions are made in the hybrid approachof CMAQv47 (1) fine particle modes are in instantaneousequilibrium with the gas phase (Capaldo et al 2000) and(2) condensation (evaporation) of HNO3 HCl and NH3 to(from) the coarse particle mode is limited such that the fluxof hydrogen ion is a maximum of 10 of the current hydro-gen ion concentration per second (Pilinis et al 2000)

The first assumption can introduce error into calculationswhen the fine modes are not in equilibrium with the gasphase However CMAQrsquos fine modes largely describe sub-micron particles with equilibration time scales comparableto those of typical gasparticle dynamics and often shorterthan an operator step of 5ndash10 min (Meng and Seinfeld 1996Dassios and Pandis 1999) The partitioning algorithm forthe fine modes involves a bulk equilibrium calculation for thecombined modes and a subsequent apportioning of mass toeach mode using weighting factors based on the modal trans-port moments (Pandis et al 1993 Binkowski and Shankar1995) Combining modes for the bulk equilibrium calcula-tion produces error when the modes have different composi-tion While this source of error may be important for finelyresolved sectional models it is not significant in CMAQwhere the overwhelming proportion of fine mass resides in asingle mode (ie the accumulation mode) Error may also beintroduced into calculations by the decoupling of interactionsbetween the gas phase and the fine and coarse modes over theoperator time step Capaldo et al (2000) reported that errordue to this decoupling became important for a 10-min opera-tor step when a large spike of NH3 was emitted during a chal-lenging portion of their box-model simulation However theerror was largely attributed to differences in particle phasestate for different decoupling times Since crystallization of

wwwgeosci-model-devnet32572010 Geosci Model Dev 3 257ndash273 2010


inorganic salts is not modeled in CMAQ CMAQrsquos aerosolcalculations are much less sensitive to decoupling time thanare calculations of the more detailed model of Capaldo etal (2000)

Pilinis et al (2000) performed sensitivity runs to evaluatethe impact of the flux limit for the hydrogen ion (ie as-sumption (2) above) They reported that varying the limiterfrom 1 sminus1 to 100 sminus1 had little impact on results To en-sure the soundness of the approach we confirmed that ourpredictions converge to the ISORROPIA equilibrium valuesafter long integration times and agree with results based on asimplified version of the Jacobson (2005) method Based onthis evidence and the success of previous studies mentionedabove the artificial flux limitation of Pilinis et al (2000) ap-pears to be a reasonable method for maintaining numericalstability while performing integrations at long time step inair quality models CMAQv47 uses a constant time step of90 s for integrating the condensation-evaporation equationsfor coarse-mode particles

212 Particle distribution geometric standarddeviations

In CMAQv46 the GSD of the coarse particle mode is fixedat 22 and sulfate is the only component to influence GSDsof the fine modes during condensation and evaporation InCMAQv47 the GSD of all three modes is variable howevera constraint is imposed such that GSDs do not change dur-ing condensation and evaporation calculations Except forthe variable GSD of the coarse mode and the condensation-evaporation constraint GSDs are calculated in CMAQv47the same way as in previous CMAQ versions (Binkowski andRoselle 2003) The constraint on GSDs during condensationand evaporation calculations is a temporary patch required toachieve stable GSD predictions and its implications are dis-cussed in Sect 42

213 Modeling chloride displacement from sea salt

In CMAQv47 HNO3 HCl and NH3 condense and evap-orate from the coarse particle mode and H2SO4 condensesThe primary advantage of the chemically-active coarse modeis that displacement of chloride by nitrate can be simulated inenvironments where sea-salt particles interact with pollutantsfrom urban areas Displacement of nitrate and chloride bysulfate is also simulated for coarse particles in CMAQv47however sulfate preferentially resides in the fine modes dueto its negligible vapor pressure and the large surface area ofthe fine modes

For solid NaCl particles exposed to HNO3at low relativehumidity (RH) the replacement of chloride by nitrate is oftenexpressed by the following heterogeneous reaction

NaCl(s)+HNO3(g) rarr NaNO3(s)+HCl(g) (R1)

(Beichert and Finlayson-Pitts 1996) However sea salt gen-erally contains highly hygroscopic salts such as calcium and

magnesium chloride in addition to sodium chloride Thesesalts have low deliquescence RHs (sim33 for MgCl26H2Oandsim28 for CaCl26H2O at 298 K compared tosim75 forNaCl) and so the mutual deliquescence RH of the sea-saltmixture should be about 30 for typical coastal conditions(eg see Figs 10ndash12 of Kelly and Wexler 2006) Also elec-trodynamic balance studies indicate that NaCl-MgCl2 andCaCl2 particles exist as supersaturated solutions at RHs wellbelow their deliquescence RH under laboratory conditions(Cohen et al 1987 Chan et al 2000) Therefore sea-saltparticles are likely to contain an aqueous electrolyte solutionat RH conditions typical of coastal environments and the dis-placement of chloride by nitrate will often occur via solutionthermodynamics rather than Reaction (R1)

Although CMAQ does not directly treat calcium or mag-nesium salts inorganic particle components are assumed toexist in aqueous solution at all RHs using the ldquometastablerdquobranch of the ISORROPIA model The pathway for ni-trate replacement of chloride in sea-salt particles in CMAQis similar to that described by Jacobson (1997) As nitricacid condenses on a sea-salt particle to maintain equilibriumwith the gas phase the particle solution concentrates Thesolution may concentrate further if the ambient RH subse-quently decreases For typical compositions the activity co-efficient of dissolved HCl increases dramatically comparedto that of dissolved HNO3 with increasing ionic strength (Ja-cobson 1997 Dasgupta et al 2007) Increases in activ-ity cause the chemical potential of dissolved HCl to exceedthat of gas-phase HCl and some HCl evaporates to main-tain equilibrium Evaporation of HCl leads to lower ionicstrength and enables nitrate to remain in solution The over-all change in particle composition for this process resemblesthat of (R1) however chloride replacement in CMAQ is re-versible and driven by solution thermodynamics rather thanbeing a kinetically-limited forward reaction

22 Parameterization of sea-salt emissions

Beginning with version 45 CMAQ has included online cal-culation of sea-salt emissions from the open ocean using themethod of Gong (2003) who extended the parameterizationof Monahan et al (1986) to submicron sizes This approachis based on the whitecap method where the emission fluxscales linearly with the fraction of ocean area covered bywhitecaps Over the open ocean whitecap coverage is de-termined as a function of wind speed using the empiricalrelation of Monahan et al (1986) The size distribution ofemitted sea salt is adjusted to local RH before mixing it withthe ambient particle modes (Zhang et al 2005)

In CMAQ primary sea-salt particles are speciated intothree components (weight by dry mass) Na+ (3856)Clminus (5389) and SO2minus

4 (755) This speciation repre-sents non-sodium sea-salt cations (eg Mg2+ Ca2+ andK+) by equivalent concentrations of sodium (on a mol basis)to achieve electroneutrality with the Clminus and SO2minus

4 anions



Moya et al (2001) demonstrate that this approach is a goodapproximation when using thermodynamic aerosol modelsthat do not include all crustal elements (eg see Fig 2 ofMoya et al 2001) To recover the sodium fraction of sea-salt cations for comparison with observations the modeledsodium concentration (ie sodium plus non-sodium sea-salt cations) is multiplied by a factor of 078 during post-processing

To account for enhanced sea-salt emission from the surfzone Nolte et al (2008) used the flux parameterizationof de Leeuw et al (2000) That treatment yielded rela-tively unbiased model results for total sodium when com-pared with observations at three BRACE sites How-ever recent improvements to the spatial allocation of surf-zone grid cells resulted in several cells close to BRACEsampling sites being reclassified as surf-zone cells (seearticle supplement (httpwwwgeosci-model-devnet32572010gmd-3-257-2010-supplementpdf) for details on surf-zone allocation) In preliminary simulations based on the deLeeuw et al (2000) parameterization with the newly grid-ded surf zone large over-predictions of sodium and chloridewere found at the coastal Azalea Park site Therefore a dif-ferent approach was needed in this study Surf-zone emis-sions are strongly dependent on local features such as waveheight and bathymetry (de Leeuw et al 2000 Lewis andSchwartz 2004) but the de Leeuw et al (2000) parame-terization was based on measurements along the Californiacoast and may not be suitable for the Florida coast For in-stance Petelski and Chomka (1996) observed significantlylower mass fluxes for the Baltic coast than were observedby de Leeuw et al (2000) for California (see discussion in deLeeuw et al 2000) However de Leeuw et al (2000) demon-strated compatibility between their surf-zone source functionand several open-ocean source functions by assuming 100whitecap coverage for the surf zone

In CMAQv47 surf-zone emission fluxes are calculatedusing the open-ocean source function of Gong (2003) witha fixed whitecap coverage of 100 and a 50-m-wide surfzone In Fig 1 this flux is compared with the surf-zonesource function of de Leeuw et al (2000) and the Clarke etal (2006) function based on 100 whitecap coverage TheClarke et al (2006) source function was developed for usein both open-ocean and coastal surf-zone environments andis based on observations of emissions from waves breakingon a Hawaiian shore All three source functions yield sim-ilar order of magnitude for a 10-m wind speed of 001 ms(Fig 1 top) however the de Leeuw et al (2000) emissionflux is much larger than the others for a 10-m wind speedof 9 ms (Fig 1 bottom) Note that the Gong (2003) andClarke et al (2006) curves do not depend on wind speed inFig 1 because the whitecap coverage is fixed Consideringthe limitations of surf-zone emission estimates (eg Lewisand Schwartz 2004 Sect 435) and the similarity of theGong (2003) flux with that derived from the surf measure-ments of Clarke et al (2006) our treatment of sea-salt emis-

SurfminusZone Flux

GongClarkede Leeuw

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=001 ms

10minus2 10minus1 100 10110minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=90 ms

Particle Diameter (μm)

dV

dD

p (

m3 μ

mminus1

mminus2

sminus1

)

Fig 1 Comparison of sea-salt emission size distributions at 80RH with 10-m wind speed (U ) of (a) 001 ms and(b) 9 ms Clarkeet al (2006) and Gong (2003) source functions are based on 100whitecap coverage the magnitude of the de Leeuw et al (2000)source function is wind-speed dependent

sion from the coastal surf-zone in CMAQv47 is reasonableHowever we will revisit this topic in the future as new ap-proaches become more established

23 Model application Tampa FL May 2002

The meteorological fields used to drive the air quality modelwere generated with the 5th generation Penn StateNCARMesoscale Model (MM5) v36 (Grell et al 1994) CMAQ-ready meteorological files were generated from the MM5simulations of Nolte et al (2008) using the Meteorology-Chemistry Interface Processor version 33 The meteorolog-ical model was configured with 30 vertical layers (11 lay-ers in the lowest 1000 m and a surface layer nominally 38 mdeep) the Pleim-Xiu planetary boundary layer and land-surface models the Grell cloud parameterization the rapidradiative transfer model and the Reisner II microphysics pa-rameterization To ensure that the simulated fields reflectedactual meteorology the model used analysis and observationnudging of temperature and moisture at the surface and aloftand of winds aloft



Table 1 Differences in CMAQ model versions used in this study

Modela Sea-salt emissionsb Coarse particlemodec

Fine-mode GSDd Coarse-mode GSDd

CMAQv46 Open-ocean only Dry chemically inert Variable influenced bycondensation of H2SO4

22

CMAQv46b Open-ocean and coastal surf-zone Dry chemically inert Variable influenced bycondensation of H2SO4

22

CMAQv46c Open-ocean and coastal surf-zone Wet dynamic masstransfer of HNO3H2SO4 HCl NH3between gas andparticle phases

Variable doesnrsquot changeduring condensation orevaporation


a CMAQv46 is a standard release version CMAQv46b and CMAQv46c are non-standard versions created for this study to evaluate theupdated sea-salt emission and coarse-particle chemistry parameterizations developed for CMAQv47b Open-ocean parameterization is that of Gong (2003) the coastal surf-zone parameterization uses the source function of Gong (2003) with100 whitecap coverage and a 50-m-wide surf zone (Sect 22)c Dynamic mass transfer is calculated using the hybrid method of Capaldo et al (2000) (Sect 211)d Particle distribution geometric standard deviations are discussed in Sect 212

Fig 2 Inner modeling domain (8 kmtimes8 km) centered on TampaFL Markers indicate land-based observational sites

An overview of CMAQ equations and algorithms is givenby Byun and Schere (2006) For our study CMAQ was con-figured to use the SAPRC99 gas-phase chemical mechanism(Carter 2000) and the Euler Backward Iterative solver Themodeling period (21 Aprilndash3 June 2002) and nested domainsmatch those of Nolte et al (2008) Specifically the outerdomain uses a 32 kmtimes32 km horizontal grid and covers thecontinental US with temporally invariant vertical concentra-tion profiles at the boundaries (Byun and Ching 1999) Theinner domain uses a 8 kmtimes8 km horizontal grid that cov-ers the Southeast US The inner domain is shown in Fig 2with markers for three BRACE observational sites Initialand boundary conditions for the inner domain were createdfrom simulations on the outer domain CMAQ-ready emis-sion files containing information on area point mobile andbiogenic sources (ie all sources except sea salt) were takenfrom Nolte et al (2008) ndash see that study for details on emis-sion inventories and uncertainty estimates

24 CMAQ model versions

Three versions of CMAQ are used in this study CMAQv46CMAQv46b and CMAQv46c CMAQv46 is a stan-dard release version and is configured as described aboveCMAQv46b is identical to CMAQv46 except that v46bincorporates the surf-zone emission parameterization devel-oped for v47 and described in Sect 22 The impact ofsurf-zone emissions of sea salt on predictions is evaluated bycomparing results of CMAQv46b with those of CMAQv46CMAQv46c is identical to CMAQv46b except that v46c in-corporates the dynamically interactive coarse particle modeand GSD treatments developed for v47 and described inSects 211 and 212 The impact of the interactive coarsemode and GSD treatments are evaluated by comparing re-sults of CMAQv46c with those of CMAQv46b CMAQv47is not used in this study because it contains numerous modelupdates in addition to those under consideration (Foley et al2010) CMAQv46b and CMAQv46c are used here to isolatethe impacts of the new treatment of sea-salt emissions and thedynamically interactive coarse particle mode These modelversions are available from the authors upon request Notethat the coarse particle mode is dry chemically inert andhas a fixed GSD of 22 in both CMAQv46 and CMAQv46bTable 1 summarizes differences of the model versions usedhere

3 Observations

CMAQ predictions are compared with observations madeat three sampling sites in the Tampa FL region (Fig 2)Azalea Park (2778 N 8274 W) Gandy Bridge (2789 N8254 W) and Sydney (2797 N 8223 W) Details on



All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

Observed Total (μg mminus3)

Pre

dic

ted

To

tal (

μg m

minus3)

Fig 3 Modeled total sodium and chloride particle concentrationsvs 23-h impactor observations at three Tampa-area sites for 5 sam-pling days (6 at Azalea Park) during 2ndash15 May 2002 ldquov46rdquo indi-cates CMAQv46 ldquov46brdquo indicates CMAQv46b see Table 1 forversion description For reference the dashed line represents 11ratio

the dataset are available in Nolte et al (2008) Arnold etal (2007) Dasgupta et al (2007) and Evans et al (2004)Briefly size-resolved measurements of inorganic PM con-centration were made with four micro-orifice cascade im-pactors which operated for 23 h per sample (Evans et al2004) Impactors had 8ndash10 fractionated stages ranging from0056 to 18 microm inDaero and two impactors were collo-cated at the Sydney site Samples were collected during23-h periods on 15 days (14 at Sydney) during 2 Mayndash2 June 2002 The sampling dates are given on figuresin the supplement (httpwwwgeosci-model-devnet32572010gmd-3-257-2010-supplementpdf) At the Sydney sitetotal (ie Daero 50 cutsim125 microm) nitrate was measuredwith 15-min resolution using a soluble particle collector andan ion chromatograph (Dasgupta et al 2007) and nitric acidwas measured continuously by denuder difference (Arnold etal 2007)

4 Results

41 Predicted and measured total PM concentrations

CMAQv46 and CMAQv46b predictions of 23-h average to-tal concentration (summed over all modes) of sodium andchloride are compared with 23-h average total observed con-centration (summed over all impactor stages) in Fig 3 forobservation days in the time period 2ndash15 May 2002 Grid-cell average predictions are compared with point measure-ments at the BRACE sites in this study The results in Fig 3demonstrate the impact of the surf-zone emission parame-terization developed for CMAQv47 When surf-zone emis-sions are neglected (ie CMAQv46) the normalized meanbias (NMB) is minus85 for sodium andminus76 for chlorideover all sites When surf-zone emissions are added to themodel (ie CMAQv46b) the sodium and chloride concen-trations increase by a factor of 28 Despite this improve-ment model predictions still fall below the observed sodiumand chloride concentrations (NMB=minus58 andminus34 forsodium and chloride respectively) This result suggests thatsea-salt emissions are significantly underestimated andor thedeposition of coarse-mode particles is too rapid in CMAQ

In Fig 4 CMAQv46b and CMAQv46c predictions of 23-h average total concentration of SO2minus

4 NH+

4 NOminus

3 Na+and Clminus are compared with 23-h average observed concen-trations at three sites for the time period 2 Mayndash2 June 2002Summary statistics for these comparisons are provided inTable 2 Differences in predictions for CMAQv46b andCMAQv46c are due to the different treatments of coarse-particle chemistry and modal GSDs described above Thelargest difference in performance between the models isfor nitrate concentration Across all sampling sites anddates nitrate is underestimated by about a factor of 10 inCMAQv46b (NMB=minus92) and only a factor of two inCMAQv46c (NMB=minus56) This substantial improvementis due to the treatment of coarse particles as chemically ac-tive in v46c but not v46b The remaining under-predictionof nitrate by CMAQv46c is comparable to that of sodium(NMB=minus56 andminus40 for nitrate and sodium respec-tively) Since sodium is the predominant cation in the coarseparticles further improvement in nitrate predictions mayrequire improvements in sea-salt emissions andor deposi-tion treatment Despite the shortcomings of the predictionsCMAQv46c estimates for total nitrate and sodium concen-tration are a clear improvement over those of CMAQv46b

The NMB and normalized mean error (NME) forCMAQv46c over all sites is improved compared toCMAQv46b for all components except chloride (Table 2 AllSites) The better performance of CMAQv46c for sodiumis perhaps surprising because sodium is non-volatile andits emissions are based on the same parameterization inv46b and v46c As explained in Sect 42 the higherpredictions of sodium concentration by CMAQv46c thanby CMAQv46b are largely due to the different treatments



Table 2 Mean observed (summed over all impactor stages) and model-predicted (summed over all modes) inorganic particle concentrations(microg mminus3) at three sites near Tampa FL

Species Obsa Modv46bb Modv46c Rv46b

c Rv46c NMBv46bd NMBv46c NMEv46b

e NMEv46c RMSEv46bf RMSEv46c

Azalea Parksulfate 403 371 382 045 045 minus79 minus53 40 39 21 21ammonium 123 093 094 051 051 minus24 minus24 33 33 06 06nitrate 196 009 081 minus007 004 minus96 minus59 96 69 20 15sodium 162 109 140 minus006 minus001 minus33 minus13 49 49 09 10chloride 193 189 198 minus004 009 minus18 25 49 57 12 13

Gandy Bridgesulfate 408 421 428 044 043 32 51 43 42 23 23ammonium 130 110 111 052 053 minus15 minus14 28 28 05 05nitrate 174 006 082 minus014 011 minus96 minus53 96 60 18 12sodium 146 054 073 052 047 minus63 minus50 63 50 11 09chloride 172 093 080 057 065 minus46 minus53 49 54 11 11

Sydneysulfate 313 259 266 047 046 minus17 minus15 30 30 12 12ammonium 104 094 095 033 034 minus88 minus80 41 41 05 05nitrate 151 030 065 minus008 040 minus80 minus57 81 60 13 10sodium 114 029 040 077 077 minus75 minus65 75 65 10 09chloride 131 049 046 077 086 minus63 minus65 63 65 10 11

All Sitessulfate 376 352 361 049 048 minus63 minus41 39 38 19 19ammonium 119 099 100 047 048 minus17 minus16 34 33 05 05nitrate 174 015 077 minus017 016 minus92 minus56 92 63 17 12sodium 141 065 086 035 034 minus54 minus40 60 54 10 09chloride 166 112 109 034 038 minus33 minus34 52 58 11 12

a Observed mean concentration (microg mminus3)b Modeled mean concentration (microg mminus3) for CMAQv46bc Pearson correlation coefficient for CMAQv46b predictionsd Normalized mean bias () for CMAQv46b predictions NMB=

sumC mod minusCobssum

Cobstimes100 whereC is concentration

e Normalized mean error () for CMAQv46b predictions NME=sum

|C mod minusCobs|sumCobs

times100

f Root mean square error (microg mminus3) for CMAQv46b predictions RMSE=

radic1

nsum

(C mod minusCobs)2 wheren is the number of samples

of GSD for the coarse particle mode The slightly higher(and better) predictions of total sulfate concentration byCMAQv46c are also attributable to the different coarse-mode GSD treatments because coarse sea-salt particles con-tain a small amount of primary sulfate (76 by dry massin CMAQ) Predictions of total ammonium concentration areessentially the same for CMAQv46b and CMAQv46c andpredictions of total chloride concentration are strongly bi-ased low for both models at the Gandy Bridge and Sydneysites (Table 2) Due to the low bias in chloride predictionsreplacement of chloride by nitrate in CMAQv46c results inslightly worse total chloride predictions for v46c than v46bat these sites However compared to standard CMAQv46which does not account for the enhanced emission of seasalt from the surf zone CMAQv46c predictions of chlo-ride concentration are an improvement (eg see Table S1in the supplementhttpwwwgeosci-model-devnet32572010gmd-3-257-2010-supplementpdf)

Comparing results across sites in Fig 4 one notices thatsodium predictions are increasingly biased low with distance

from the Gulf of Mexico Error in transport and depositionof sea-salt particles from the gulf could be responsible forthis behavior A related possibility is that relatively fine-scalecoastal processes are not adequately captured with the 8-kmhorizontal resolution used in this study Also error in sea-salt emissions from the bay which are calculated accordingto the open-ocean algorithm could potentially lead to spa-tial differences in performance For instance bay emissionswould impact the Gandy Bridge site most due to its baysidelocation (Fig 2) and would influence the Sydney and AzaleaPark sites differently for flows to and away from the gulf

Overall results in Fig 4 and Table 2 indicate that thedynamically interactive coarse particle mode developed forCMAQv47 greatly improves predictions of total nitrate con-centration and slightly improves predictions of total sulfateammonium and sodium concentration near the coast Re-sults in Fig 3 and Table S1 indicate that the surf-zone emis-sion parameterization developed for CMAQv47 improvespredictions of total sodium and chloride concentration nearthe coast



0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)

Azalea Park Gandy Bridge Sydney All Sites

Fig 4 Modeled total inorganic particle concentrations vs 23-h impactor observations at three Tampa-area sites for 15 sampling days (14at Sydney) during 2 Mayndash2 June 2002 ldquov46brdquo indicates CMAQv46b ldquov46crdquo indicates CMAQv46c see Table 1 for version descriptionFor reference the dashed line represents 11 ratio See Table 2 for summary statistics

42 Predicted and measured particle size distributions

Size distributions of SO2minus

4 NH+

4 NOminus

3 Na+ and Clminus pre-dicted by CMAQv46b and CMAQv46c are compared withspeciated impactor measurements averaged over all sampling

days in Fig 5 Modeled diameters were converted toDaerofor comparison with the impactor data Since the four im-pactors did not have identical size cuts observations wereaveraged to the size grid of a lower-resolution (8 fractionated



0005101520253035 SO4

2minusObservedCMAQv46bCMAQv46c

0002040608101214

NH4+

ObservedCMAQv46bCMAQv46c

0002040608101214

NO3minus


0002040608101214

Na+ObservedCMAQv46bCMAQv46c

05 2 1 5 200002040608101214

ClminusObservedCMAQv46bCMAQv46c

SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20


Aerodynamic diameter (μm)

dM

(μg

mminus3

)d

lnD

p

Azalea Park Gandy Bridge Sydney

Fig 5 Observed and predicted size distributions of inorganic particle components at three Tampa-area sites averaged over 15 sampling days(14 at Sydney) during 2 Mayndash2 June 2002 Vertical dashed line indicatesDaeroof 25 microm

stages) impactor for this figure A figure similar to Fig 5 butwith CMAQ distributions mapped to the 8-stage size grid isgiven in the supplement (Fig S1) Comparisons of modelpredictions with observations at the original impactor reso-lutions for individual sampling days are also available in thesupplementary material (Figs S2ndashS15)

Both CMAQv46b and CMAQv46c correctly predict thatammonium and sulfate reside predominantly in fine particles(see top two panels of Fig 5) CMAQv46b predicts higherdistribution peaks for these species than does CMAQv46cThis difference is due in part to differences in the treatmentsof GSDs for the particle distributions CMAQv46b allows



fine mode GSDs to vary during sulfate condensation cal-culations whereas CMAQv46c does not Condensationalgrowth narrows a size distribution because the diameters ofsmall particles increase relatively quickly compared to thoseof large particles due to the higher surface area-to-volumeratios of small particles Therefore CMAQv46b predictsslightly narrower fine particle modes and higher size distri-bution peaks than does CMAQv46c the average GSD ofthe accumulation mode is 202 for CMAQv46b and 205 forCMAQv46c over all sites and sampling days Another po-tential reason for higher peaks in the ammonium and sul-fate distributions of CMAQv46b is that small amounts ofammonia and sulfuric acid condense on coarse particles inCMAQv46c reducing their availability for condensation onfine particles However the mass of ammonium in the coarsemode is on average only 3 of that in the fine modes andso uptake of ammonia by the coarse mode does not signif-icantly impact the fine particle distribution Similarly themass of sulfate in the coarse mode is small and due in part toprimary emissions of sulfate in coarse sea-salt particles

The largest difference in the size-distribution predictionsof CMAQv46b and CMAQv46c is for nitrate The chem-ically active coarse mode enables CMAQv46c to correctlypredict that nitrate predominantly resides in coarse parti-cles (Fig 5) CMAQv46b does not allow the formationof coarse particle nitrate and cannot realistically simulatethe nitrate size distribution at these three coastal observationsites Despite the better performance of CMAQv46c for ni-trate under-prediction of sodium the primary coarse parti-cle cation leads to under-prediction of coarse nitrate At theSydney site the models under-predict sodium and nitrate inthe coarse mode and over-predict nitrate in the accumulationmode (Fig 5) The prediction of significant accumulation-mode nitrate at Sydney (but not the other sites) is primar-ily due to concentrations of ammonia in excess of those re-quired to fully neutralize aqueous sulfuric acid at SydneyThe average predicted molar ratio of total ammonia (ieNH3+NH+

4 ) to non-sea-salt SO2minus

4 is greater than four at theSydney site and only about two at Azalea Park and GandyBridge Ammonia ratios greater than two exceed that of neu-tral (NH4)2SO4 and facilitate nitric acid condensation by en-abling significant neutralization of aqueous nitric acid Over-prediction of accumulation-mode nitrate is more pronouncedfor CMAQv46b than CMAQv46c because CMAQv46bdoes not have a pathway for coarse-mode nitrate formation

Both models correctly predict that sodium and chloride re-side predominantly in coarse particles (see bottom two panelsof Fig 5) However CMAQv46c predicts higher concentra-tions of sodium than does CMAQv46b in better agreementwith the measurements Averaged over all sites and samplingdays the sodium concentration predicted by CMAQv46cis 32 greater than that predicted by CMAQv46b Sinceemissions of sea salt are based on the same parameteriza-tion in CMAQv46b and CMAQv46c differences in sodiumpredictions are attributable to differences in advective trans-

Fraction of Total Nitrate in Particle Phase

01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08

10 Observed CMAQv46b CMAQv46c

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 01 02Day (May 17 minus June 2 2002)

00

02

04

06

08

10

Fig 6 Time series of observed and modeled fraction of total nitrate

in the particle phase [ie NOminus3

(HNO3+NOminus

3 )] at the Sydney

FL site from 1 Mayndash2 June 2002 Tick marks represent 0000 localstandard time on each day

port and deposition The combined effect of these pro-cesses differs for the models largely because CMAQv46buses a fixed GSD of 22 for the coarse particle mode whileCMAQv46c uses a variable coarse-mode GSD which hasan average value of 206 during the observation period Thelower coarse-mode GSD for CMAQv46c appears to resultin lower dry deposition and in better predictions of coarsesodium concentration by v46c than v46b

Both models over-predict the geometric mean diameter(GMD) of the accumulation mode (Figs 5 and S1) Over-prediction of GMD also occurs for the coarse mode (seeFigs S2ndashS15 for individual days) however this behavior isnot evident in Fig 5 because the impactor measurementshave been averaged to an 8-stage size distribution Over-prediction of GMD could cause over-prediction of dry de-position and increasing low bias of concentration predictionswith distance from a source The peaks in the observed sizedistributions of sulfate and ammonium occur in the size binwith GMD of 040 microm For CMAQ distributions that havebeen mapped to the impactor size grid (Fig S1) the mod-eled peaks for sulfate and ammonium occur in the adjacentlarger bin which has a GMD of 075 microm Although thisdifference could suggest an over-prediction of accumulationmode GMD of about 035 microm or 88 by CMAQ the exactover-prediction cannot be quantified due to the limited im-pactor resolution and the different representations of the par-ticle size distribution by CMAQ and the cascade impactorSimilarly GSD of the accumulation mode appears to be over-predicted by CMAQ based on visual inspection of Figs 5 andS1 but the exact over-prediction cannot be reliably quanti-fied



00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io


Fig 7 Observed and modeled molar ratios of average inorganic ion concentration to average sodium ion concentration at three Tampa-area

sites for 15 sampling days (14 at Sydney) during 2 Mayndash2 June 2002 Horizontal dashed lines indicate average Clminus

Na+ and SO2minus

4

Na+

ratios in seawater

Since fine and coarse particles have different sources theover-prediction of GMD is not easily attributable to an in-correct emission size distribution Modal GMD is diagnosedfrom the zeroth second and third moments of the particlesize distribution in CMAQ and so the cause of the diame-ter over-prediction is not obvious Zhang et al (2006) re-ported similar over-prediction of volume mean diameter byCMAQ for a site in Atlanta in summer and Elleman andCovert (2009) reported that CMAQ size distributions areshifted to larger sizes compared with observations at Lan-gley Bristish Columbia in August Therefore the problem

of diameter over-prediction is not confined to conditions ofthe BRACE campaign Note that PM25 predictions wouldincrease slightly if over-predictions of GMD were correctedbecause a larger fraction of the accumulation mode wouldfall below 25 microm (Jiang et al 2006) Also note that predic-tions ofDaero for coarse particle modes by CMAQv46b andCMAQv46c are similar even though the coarse mode doesnot contain water in v46b In the calculation ofDaero therelatively low density of water compared to that of dry sea-salt components compensates for the larger Stokes diameterspredicted by CMAQv46c



43 Predictions and measurements of nitratepartitioning

Predictions of the mass fraction of nitrate in the parti-cle phase [ie NOminus3

(HNO3+NOminus

3 )] are compared withhighly time-resolved measurements made at the Sydney sitein Fig 6 The average value of the particle fraction of nitrateover the observation period is 051 for the measurements035 for CMAQv46c and 013 for CMAQv46b Thereforethe chemically-active coarse particle mode greatly improvespredictions of nitrate partitioning by CMAQ Despite this im-provement CMAQv46c generally under-predicts the parti-cle fraction of nitrate Also although the timing of manypeaks in the observed time series is correctly predicted thediurnal amplitude of the measurements is not adequately cap-tured by the model However CMAQv46c is a clear im-provement over CMAQv46b which incorrectly predicts thatthe particle fraction of nitrate is negligible for many time pe-riods CMAQv46c also provides better predictions of ab-solute concentrations of HNO3 and HCl than CMAQv46b(eg see Figs S16 and S17 inhttpwwwgeosci-model-devnet32572010gmd-3-257-2010-supplementpdf)

The under-prediction of the fraction of nitrate in the par-ticle phase by CMAQv46c could be due to the under-prediction of sodium ion discussed above To investigate thispossibility the average molar ratios of the inorganic ions tothe sodium ion are examined for the two highest fraction-ated stages (18ndash32 and 32ndash18 microm Fig 7) CMAQ predic-tions were mapped to these stages by integrating the distribu-tions in Fig 5 over the impactor size ranges The measuredammonium-to-sodium ratios are negligible for these stagesand suggest that sodium is the dominant cation forDaerogt

32 microm In contrast to the observations both CMAQv46band CMAQv46c predict amounts of ammonium and sulfatecomparable to that of sodium in the lower of the two size bins(Fig 7 bottom two rows) This behavior is attributable tothe over-prediction of GMD and possibly GSD by CMAQ(Figs 5 and S1) The error in CMAQv46c predictions of themolar ratios of nitrate and chloride to sodium for the lowerstage (Fig 7 top two rows) may reflect a limitation of usinga single mode to represent all coarse particles

Since the models correctly predict that the ammonium-to-sodium ratios are negligibly small for the highest stagethe influence of sodium on nitrate partitioning predictionscan be evaluated by focusing on this stage If under-prediction of nitrate is primarily a consequence of under-prediction of sodium the nitrate-to-sodium ratios shouldbe in reasonable agreement with the observations For theGandy Bridge and Sydney sites CMAQv46c predictions ofthe nitrate-to-sodium ratio agree well with observations de-spite the large under-prediction of absolute nitrate concen-tration The nitrate-to-sodium molar ratio is under-predictedby CMAQv46c by only 05 at Gandy Bridge and by only75 at Sydney whereas absolute nitrate concentration isunder-predicted by 53 at Gandy Bridge and 57 at Syd-

ney The molar ratios of the other inorganic ions are alsoin reasonable agreement with measurements at these sitesTherefore the under-prediction of nitrate and particle frac-tion of nitrate by CMAQv46c is largely attributable to theunder-prediction of sodium ion This finding suggests thatthe dynamically interactive coarse particle mode is function-ing properly but that emissions transport and depositionof sodium are not adequately captured by the model for theTampa domain In contrast to the good predictions for theGandy Bridge and Sydney sites the nitrate-to-sodium mo-lar ratio is under-predicted by 49 by CMAQv46c at Aza-lea Park The Azalea Park site is located in a grid cell withsurf-zone emissions of sea salt and so the error in the mod-eled nitrate-to-sodium ratio at this site may reflect the poorrepresentation of the mixing of marine and continental airmasses in the grid cell However the good predictions ofthe nitrate-to-sodium ratio at the inland (Sydney) and non-surf-zone bay site (Gandy Bridge) indicate that the sea-saltchemical-processing calculations are reliable

44 Model timing

Computational efficiency is a key aspect of the model devel-opments described here Models that are significantly slowerthan CMAQv46 are not suitable for conducting the numer-ous long-term simulations required for developing State Im-plementation Plans for the annual PM25 standard The runtime of CMAQv46c is only about 8 longer than that ofCMAQv46 This increase is modest considering the sig-nificantly better predictions of CMAQv46c at the coastalBRACE sites The primary cause of the longer run time forCMAQv46c is the additional calls to the ISORROPIA ther-modynamic module used in simulating dynamic mass trans-fer of coarse-particle components in CMAQv46c

5 Closing remarks

This study focuses on evaluating parameterizations of sea-salt emissions from the coastal surf zone and the dynamictransfer of HNO3 H2SO4 HCl and NH3 between coarseparticles and the gas phase in CMAQ The methods describedabove improve predictions of inorganic particle componentsand nitrate partitioning at sites near Tampa Bay FL and areincluded in the public release of CMAQ version 47 Whilethe updates to CMAQ clearly improve predictions for condi-tions of the BRACE campaign several areas for future modeldevelopment were identified

First particle size distributions from CMAQ do not ade-quately capture the narrow distribution peaks of the obser-vations The opposite problem (ie modeled distributionstoo narrow) was reported by Nolte et al (2008) for a sim-ulation of the same domain with the CMAQ-UCD modelThe causes of this difference should be determined in a fu-ture study Second particle-mode GMDs are over-predicted



by CMAQ Considering that Zhang et al (2006) and Elle-man and Covert (2009) also report over-prediction of diame-ter by CMAQ and that this discrepancy may influence PM25predictions the source of the error should be investigated infuture work

Another area for future model development is on improv-ing the simulation of sea-salt emissions from the coastal surfzone The surf-zone emission parameterization developedfor CMAQv47 improves predictions of sodium and chlorideconcentration in the Tampa area Yet predictions of sodiumare increasingly biased low with distance from the Gulf ofMexico This behavior was not observed in the Nolte etal (2008) study and could be due to inadequate sea-salt emis-sions in addition to the over-predictions of GMD and GSDwhich could yield too rapid particle deposition rates TheClarke et al (2006) parameterization (Fig 1) produces highersea-salt emissions than the modified Gong (2003) functionused in CMAQv47 and could improve predictions for theTampa domain However emissions of sea salt from the surfzone are dependent on local features and the ideal parame-terization for Tampa may not be suitable for other locationswhere CMAQ is applied Possibly a parameterization couldbe developed that adapts to local features or multiple param-eterizations could be incorporated into CMAQ and appliedseparately in different parts of the domain

In addition to the emission parameterization error insodium and chloride predictions can be attributed to using8 kmtimes8 km horizontal grid cells for simulating relativelyfine-scale coastal processes Athanasopoulou et al (2008)recently used 2 kmtimes2 km horizontal grid cells in a nestedportion of their domain to capture fine-scale processes nearthe coast Predictions were not evaluated quantitatively inthat study because measurements are not available duringthe simulation period Using higher grid resolution and tun-ing sea-salt emission from the surf zone could result in betterpredictions of the BRACE observations However the goalof our development is a model that can be applied generallyby CMAQ users who are often constrained to coarse gridresolutions and do not focus on the Tampa area

While the model updates are evaluated here for condi-tions of Tampa a separate study (Foley et al 2010) sug-gests that the updates improve model performance in severalcoastal environments In that study CMAQv47 simulationswith and without the new model features are performed forthe eastern US with 12-km horizontal resolution and pre-dictions are compared with observations from nine coastalCASTNET (Clarke et al 1997) and four coastal SEARCH(Hansen et al 2003) sites For the CASTNET sites theupdated sea-salt emissions and coarse particle processes de-crease the mean absolute error (MAE) for nitric acid predic-tions by 36 in January and by 33 in August 2006 whileMAE for total particle nitrate decreases by 10 in Januaryand by 1 in August 2006 For the SEARCH sites themodel updates decrease MAE for coarse particle nitrate by45 in January and by 52 in August 2006 while MAE for

fine particle nitrate decreases by 05 in January and by 11in August 2006 These simulations are thoroughly discussedby Foley et al (2010) The comparisons with coastal CAST-NET and SEARCH observations build confidence that themodeling approaches described here improve CMAQ predic-tions across a range of coastal conditions However accurateprediction of fine-scale coastal processes probably requiresusing higher grid resolution and a surf-zone emission param-eterization tailored to local conditions

AcknowledgementsWe kindly thank the following individuals fortheir assistance in conducting this study W Benjey R DennisA Eyth E Kinnee R Mathur S Pandis T Pierce G PouliotS Roselle K Sartelet K Schere and M Wilson

Disclaimer The United States Environmental ProtectionAgency through its Office of Research and Development fundedand managed the research described here It has been subjectedto the Agencyrsquos administrative review and approved for publication

Edited by O Boucher

References

Allen A G Harrison R M and Erisman J W Field-measurements of the dissociation of ammonium-nitrate andammonium-chloride aerosols Atmos Environ 23(7) 1591ndash1599 1989

Arnold J R Hartsell B E Luke W T Ullah S M R Das-gupta P K Huey L G and Tate P Field test of four meth-ods for gas-phase ambient nitric acid Atmos Environ 41(20)4210ndash4226 2007

Asgharian B Hoffman W and Bergmann R Particle deposi-tion in a multiple-path model of the human lung Aerosol SciTechnol 34 332ndash339 2001

Athanasopoulou E Tombrou M Pandis S N and Russell AG The role of sea-salt emissions and heterogeneous chemistryin the air quality of polluted coastal areas Atmos Chem Phys8 5755ndash5769 2008httpwwwatmos-chem-physnet857552008

Atkeson T Greening H and Poor N Bay Region At-mospheric Chemistry Experiment (BRACE) Atmos Environ41(20) 4163ndash4164 2007

Beichert P and Finlayson-Pitts B J Knudsen cell studies ofthe uptake of gaseous HNO3 and other oxides of nitrogen onsolid NaCl The role of surface-adsorbed water J Phys Chem100(37) 15218ndash15228 1996

Bessagnet B Hodzic A Vautard R Beekmann M CheinetS Honore C Liousse C and Rouil L Aerosol modelingwith CHIMERE ndash preliminary evaluation at the continental scaleAtmos Environ 38(18) 2803ndash2817 2004

Binkowski F S and Roselle S J Models-3 community multi-scale air quality (CMAQ) model aerosol component ndash 1 Modeldescription J Geophys Res 108(D6) 4183ndash4201 2003

Bricker S Longstaff B Dennison W Jones A Boicourt KWicks C and Woerner J Effects of Nutrient Enrichment In theNationrsquos Estuaries A Decade of Change NOAA Coastal Ocean



Program Decision Analysis Series No 26 National Centers forCoastal Ocean Science Silver Spring MD 2007

Binkowski F and Shankar U The Regional Particulate MatterModel 1 Model description and preliminary results J GeophysRes 100(D12) 26191ndash26209 1995

Brunekreef B and Forsberg B Epidemiological evidence of ef-fects of coarse airborne particles on health European RespiratoryJ 26(2) 309ndash318 2005

Byun D and Ching J K S Science algorithms of the EPAModels-3 Community Multiscale Air Quality (CMAQ) modelingsystem Tech Rep EPA-600R-99030 US Government PrintingOffice US Environmental Protection Agency Washington DC1999

Byun D and Schere K L Review of the governing equationscomputational algorithms and other components of the Models-3 Community Mulitiscale Air Quality (CMAQ) modeling sys-tem Appl Mech Rev 59 51ndash77 2006

Capaldo K P Pilinis C and Pandis S N A computationallyefficient hybrid approach for dynamic gasaerosol transfer in airquality models Atmos Environ 34(21) 3617ndash3627 2000

Carter W P L Implementation of the SAPRC-99 chemicalmechanism into the Models-3 framework Tech Rep US En-vironmental Protection Agency URLhttpwwwcertucredusimcarterabstshtms99mod3 2000

Chan C K Ha Z Y and Choi M Y Study of water activitiesof aerosols of mixtures of sodium and magnesium salts AtmosEnviron 34(28) 4795ndash4803 2000

Clarke J F Edgerton E S and Martin B E Dry depositioncalculations for the clean air status and trends network AtmosEnviron 31(21) 3667ndash3678 1997

Clarke A D Owens S R and Zhou J C An ultrafine sea-saltflux from breaking waves Implications for cloud condensationnuclei in the remote marine atmosphere J Geophys Res 111D06202 doi1010292005JD006565 2006

Cohen M D Flagan R C and Seinfeld J H Studies of concen-trated electrolyte-solutions using the electrodynamic balance1Water activities for single-electrolyte solutions J Phys Chem91(17) 4563ndash4574 1987

Dasgupta P K Campbell S W Al-Horr R S Ullah S M RLi J Z Amalfitano C and Poor N D Conversion of sea saltaerosol to NaNO3 and the production of HCl analysis of tempo-ral behavior of aerosol chloridenitrate and gaseous HClHNO3Atmos Environ 41(20) 4242ndash4257 2007

Dassios K G and Pandis S N The mass accommodation co-efficient of ammonium nitrate aerosol Atmos Environ 33(18)2993ndash3003 1999

de Leeuw G Neele F P Hill M Smith M H and VignatiE Production of sea spray aerosol in the surf zone J GeophysRes 105(D24) 29397ndash29409 2000

Evans M S C Campbell S W Bhethanabotla V and Poor ND Effect of sea salt and calcium carbonate interactions withnitric acid on the direct dry deposition of nitrogen to Tampa BayFlorida Atmos Environ 38(29) 4847ndash4858 2004

Elleman R A and Covert D S Aerosol size distributionmodeling with the Community Multiscale Air Quality mod-eling system (CMAQ) in the Pacific Northwest 1 Modelcomparison to observations J Geophys Res 114 D11206doi1010292008JD010791 2009

Foley K M Roselle S J Appel K W Bhave P V Pleim J

E Otte T L Mathur R Sarwar G Young J O GilliamR C Nolte C G Kelly J T Gilliland A B and Bash JO Incremental testing of the Community Multiscale Air Quality(CMAQ) modeling system version 47 Geosci Model Dev 3205ndash226 2010

Foltescu V L Pryor S C and Bennet C Sea salt generationdispersion and removal on the regional scale Atmos Environ39(11) 2123ndash2133 2005

Gard E E Kleeman M J Gross D S Hughes L S Allen JO Morrical B D Fergenson D P Dienes T Galli M EJohnson R J Cass G R and Prather K A Direct observa-tion of heterogeneous chemistry in the atmosphere Science 2791184ndash1187 1998

Gaydos T M Koo B Pandis S N and Chock D P De-velopment and application of an efficient moving sectional ap-proach for the solution of the atmospheric aerosol conden-sationevaporation equations Atmos Environ 37(23) 3303ndash3316 2003

Gong S L A parameterization of sea-salt aerosol source functionfor sub- and super-micron particles Global Biogeochem Cycles17(4) 1097 doi1010292003GB002079 2003

Grell G Dudhia J and Stauffer D A description of the fifth-generation Penn StateNCAR mesoscale model (MM5) TechRep NCARTN-398+STR National Center for Atmospheric Re-search Boulder CO 1994

Grell G A Peckham S E Schmitz R McKeen S A Frost GSkamarock W C and Eder B Fully coupled ldquoonlinerdquo chem-istry within the WRF model Atmos Environ 39(37) 6957ndash6975 2005

Hansen D A Edgerton E S Hartsell B E Jansen J J Kan-dasamy N Hidy G M and Blanchard C L The southeasternaerosol research and characterization study Part 1-Overview JAir Waste Manag Assoc 53(12) 1460ndash1471 2003

Hopkins R J Desyaterik Y Tivanski A V Zaveri R ABerkowitz C M Tyliszczak T Gilles M K and Laskin AChemical speciation of sulfur in marine cloud droplets and parti-cles Analysis of individual particles from the marine boundarylayer over the California current J Geophys Res 113 D04209doi1010292007JD008954 2008

Hsu S-C Liu S C Kao S-J Jeng W-L Huang Y-T TsengC-M Tsai F Tu J-Y and Yang Y Water-soluble speciesin the marine aerosol from the northern South China Sea Highchloride depletion related to air pollution J Geophys Res 112D19304 doi1010292007JD008844 2007

Jacobson M Z Development and application of a new air pollu-tion modeling system3 Aerosol-phase simulations Atmos En-viron 31(4) 587ndash608 1997

Jacobson M Z A solution to the problem of nonequilibriumacidbase gas-particle transfer at long time step Aerosol SciTechnol 39(2) 92ndash103 2005

Jiang W Smyth S GirouxE Roth H and Yin D Differencesbetween CMAQ fine mode particle and PM25 concentrationsand their impact on model performance evaluation in the lowerFraser valley Atmos Environ 40(26) 4973ndash4985 2006

Keene W C Stutz J Pszenny A A P Maben J R FischerE V Smith A M von Glasow R Pechtl S Sive B Cand Varner R K Inorganic chlorine and bromine in coastalNew England air during summer J Geophys Res 112 D10S12doi1010292006JD007689 2007



Kelly J T and Wexler A S Water uptake by aerosol Water ac-tivity in supersaturated potassium solutions and deliquescence asa function of temperature Atmos Environ 40(24) 4450ndash44682006

Kleeman M J and Cass G R A 3D Eulerian source-orientedmodel for an externally mixed aerosol Environ Sci Technol35 4834ndash4848 2001

Koo B Gaydos T M and Pandis S N Evaluation of theequilibrium dynamic and hybrid aerosol modeling approachesAerosol Sci Technol 37(1) 53ndash64 2003

Lee T Yu X Y Ayres B Kreidenweis S M Malm W Cand Collett J L Observations of fine and coarse particle nitrateat several rural locations in the United States Atmos Environ42(11) 2720ndash2732 2008

Lewis E R and Schwartz S E Sea-Salt Aerosol ProductionMechanisms Methods Measurements and Models ndash A CriticalReview American Geophysical Union Washington DC 2004

Lurmann F W Wexler A S Pandis S N Musarra S KumarN and Seinfeld J H Modelling urban and regional aerosols2Application to Californiarsquos South Coast Air Basin Atmos Envi-ron 31(17) 2695ndash2715 1997

McInnes L M Covert D S Quinn P K and Germani MS Measurements of chloride depletion and sulfur enrichmentin individual sea-salt particles collected from the remote marineboundary-layer J Geophys 99(D4) 8257ndash8268 1994

Meng Z Y Dabdub D and Seinfeld J H Size-resolved andchemically resolved model of atmospheric aerosol dynamics JGeophys Res 103(D3) 3419ndash3435 1998

Meng Z Y and Seinfeld J H Time scales to achieve atmosphericgas-aerosol equilibrium for volatile species Atmos Environ30(16) 2889ndash2900 1996

Monahan E C Spiel D E and Davidson K L A model ofmarine aerosol generation via whitecaps and wave disruption inOceanic Whitecaps edited by Monahan E C and MacNiocaillG D Reidel Publishing Company Norwell Mass pp 167ndash174 1986

Moya M Ansari A S and Pandis S N Partitioning of nitrateand ammonium between the gas and particulate phases duringthe 1997 IMADA-AVER study in Mexico City Atmos Environ35 1791ndash1804 2001

Nenes A Pandis S N and Pilinis C ISORROPIA A new ther-modynamic equilibrium model for multiphase multicomponentinorganic aerosols Aquatic Geochem 4(1) 123ndash152 1998

Nicholls R J and Small C Improved estimates of coastal popu-lation and exposure to hazards released EOS Transactions July83(28) 301 amp 305 2002

Nolte C G Bhave P V Arnold J R Dennis R L Zhang KM and Wexler A S Modeling urban and regional aerosolsndash Application of the CMAQ-UCD Aerosol Model to Tampa acoastal urban site Atmos Environ 42(13) 3179ndash3191 2008

Osthoff H D Roberts J M Ravishankara A R Williams E JLerner B M Sommariva R Bates T S Coffman D QuinnP K Dibb J E Stark H Burkholder J B Talukdar R KMeagher J Fehsenfeld F C and Brown S S High levels ofnitryl chloride in the polluted subtropical marine boundary layerNature Geosci 1(5) 324ndash328 2008

Pandis S N Wexler A S and Seinfeld J H Secondary or-ganic aerosol formation and transport2 predicting the ambi-ent secondary organic aerosol-size distribution Atmos Environ

27(15) 2403ndash2416 1993Park S K Marmur A Kim S B Tian D Hu Y T McMurry

P H and Russell A G Evaluation of fine particle numberconcentrations in CMAQ Aerosol Sci Technol 40(11) 985ndash996 2006

Petelski T and Chomka M Marine aerosol fluxes in the coastalzonendashBAEX experimental data Oceanologia 38 469ndash4841996

Pierce J R and Adams P J Global evaluation of CCN formationby direct emission of sea salt and growth of ultrafine sea salt JGeophys Res 111 D06203 doi1010292005JD006186 2006

Pilinis C Capaldo K P Nenes A and Pandis S N MADMndash A new multicomponent aerosol dynamics model Aerosol SciTechnol 32(5) 482ndash502 2000

Pleim J Wong D Mathur R Young J Otte T Gilliam RBinkowski F and Xiu A Development of the coupled 2-wayWRF-CMAQ system 7th Annual CMAS Conference 6ndash8 Octo-ber Chapel Hill NChttpwwwcmascenterorg 2008

Pryor S C Barthelmie R J Schoof J T Binkowski F S DelleMonache L and Stull R Modeling the impact of sea-spray onparticle concentrations in a coastal city Sci Tot Environ 391132ndash142 2008

Pryor S C and Sorensen L L Nitric acid-sea salt reactions Im-plications for nitrogen deposition to water surfaces J Appl Me-teorol 39(5) 725ndash731 2000

Sandstrom T Nowak D and van Bree L Health effectsof coarse particles in ambient air messages for research anddecision-making European Respiratory J 26(2) 187ndash1882005

Sartelet K N Hayami H Albriet B and Sportisse B De-velopment and preliminary validation of a modal aerosol modelfor tropospheric chemistry MAM Aerosol Sci Technol 40(2)118ndash127 2006

Sartelet K N Hayami H and Sportisse B Dominant aerosolprocesses during high-pollution episodes over Greater Tokyo JGeophys Res 112 D14214 doi1010292006JD007885 2007

Sarwar G and Bhave P V Modeling the effect of chlorine emis-sions on ozone levels over the eastern United States J ApplMeteorol Climatol 46(7) 1009ndash1019 2007

Seinfeld J H and Pandis S N Atmospheric Chemistry andPhysics From Air Pollution to Climate Change John Wiley ampSons Inc New York 1998

Simon H Kimura Y McGaughey G Allen D T Brown S SOsthoff H D Roberts J M Byun D and Lee D Modelingthe impact of ClNO2 on ozone formation in the Houston areaJ Geophys Res 114 D00F03 doi1010292008JD0107322009

Smyth S C Jiang W M Roth H Moran M D Makar PA Yang F Q Bouchet V S and Landry H A comparativeperformance evaluation of the AURAMS and CMAQ air-qualitymodelling systems Atmos Environ 43(5) 1059ndash1070 2009

Spyridaki A Lazaridis M Eleftheriadis K Smolik J Mi-halopoulos N and Aleksandropoulou V Modelling and eval-uation of size-resolved aerosol characteristics in the easternMediterranean during the SUB-AERO project Atmos Environ40 6261ndash6275 2006

Sullivan R C and Prather K A Investigations of the diurnalcycle and mixing state of oxalic acid in individual particles inAsian aerosol outflow Environ Sci Technol 41(23) 8062ndash



8069 2007Sun Q and Wexler A S Modeling urban and regional aerosols ndash

Condensation and evaporation near acid neutrality Atmos Env-iron 32(20) 3527ndash3531 1998a

Sun Q and Wexler A S Modeling urban and regional aerosolsnear acid neutrality ndash Application to the 24ndash25 June SCAQSepisode Atmos Environ 32(20) 3533ndash3545 1998b

Tang I Tridico A and Fung K Thermodynamic and opti-cal properties of sea salt aerosols J Geophys Res 102(D19)23269ndash23275 1997

Volckens J Dailey L Walters G and Devlin R B Di-rect particle-to-cell deposition of coarse ambient particulate mat-ter increases the production of inflammatory mediators fromcultured human airway epithelial cells Environ Sci Technol43(12) 4595ndash4599 2009

Wexler A S and Seinfeld J H The distribution of ammonium-salts among a size and composition dispersed aerosol AtmosEnviron 24(5) 1231ndash1246 1990

Wexler A S and Seinfeld J H Analysis of aerosol ammonium-nitrate ndash departures from equilibrium during SCAQS Atmos En-viron 26(4) 579ndash591 1992

Zaveri R A Easter R C Fast J D and Peters L K Model forSimulating Aerosol Interactions and Chemistry (MOSAIC) JGeophys Res 113 D13204 doi1010292007JD008782 2008

Zhang K M Knipping E M Wexler A S Bhave P V andTonnesen G S Size distribution of sea-salt emissions as a func-tion of relative humidity Atmos Environ 39(18) 3373ndash33792005

Zhang K M and Wexler A S An asynchronous time-stepping(ATS) integrator for atmospheric applications Aerosol dynam-ics Atmos Environ 40(24) 4574ndash4588 2006

Zhang K M and Wexler A S Modeling urban and regionalaerosols ndash Development of the UCD Aerosol Module and im-plementation in CMAQ model Atmos Environ 42(13) 3166ndash3178 2008

Zhang Y Liu P Pun B and Seigneur C A comprehensiveperformance evaluation of MM5-CMAQ for the summer 1999southern oxidants study episode ndash Part III diagnostic and mech-anistic evaluations Atmos Environ 40(26) 4856ndash4873 2006

Zhang Y Pun B Vijayaraghavan K Wu S Y Seigneur CPandis S N Jacobson M Z Nenes A and Seinfeld J HDevelopment and application of the model of aerosol dynam-ics reaction ionization and dissolution (MADRID) J GeophysRes 109 D01202 doi1010292003JD003501 2004



large contribution of sea salt to atmospheric particulate mat-ter (PM) the emission and chemical evolution of sea-salt par-ticles must be represented accurately by models

The diameter of sea-salt particles spans several orders ofmagnitude but the peak in the mass distribution is usu-ally in the coarse size range (aerodynamic diameterDaerogt25 microm) (eg Keene et al 2007) Uptake of gaseousspecies by coarse sea-salt particles reduces their availabilityfor condensation on fine particles and can potentially reducethe mass concentration of PM25 (PM with Daerole 25 microm)Uptake by coarse sea salt can also significantly reduce theconcentration of nitric acid in environments where the for-mation of particulate ammonium nitrate is unfavorable (egammonia-limited or high-temperature) Associations be-tween coarse particle nitrate and sea salt have been observedin both coastal (eg Hsu et al 2007) and rural (eg Lee etal 2008) areas

Sea-salt emissions are enhanced in the coastal surf zonecompared to the open ocean and result in elevated concentra-tions near the coast (de Leeuw et al 2000) During advec-tion toward land sea salt is often exposed to anthropogenicemissions from shipping lanes (Osthoff et al 2008 Simonet al 2009) and coastal urban centers (Nolte et al 2008)Considering that many coastal areas are densely populated(Nicholls and Small 2002) chemical modification of sea-salt particles by acidic gases could result in significant humanexposure to anthropogenic PM10 (PM with Daerole 10 microm) incoastal environments This exposure is a concern in light ofassociations between increases in coarse particle concentra-tions and adverse health effects (Brunekreef and Forsberg2005 Sandstrom et al 2005 Volckens et al 2009)

Despite the significance of sea-salt emissions and chemi-cal transformations some prominent air quality models havenot treated sea-salt particles (eg Bessagnet et al 2004Grell et al 2005) Other models have included emissionsof sea-salt particles but have not simulated their chemi-cal interactions with gas-phase species (eg Foltescu et al2005 Smyth et al 2009) The US EPArsquos Community Mul-tiscale Air Quality (CMAQ) model has included online cal-culation of sea-salt emissions from the open ocean since ver-sion 45 but has not accounted for enhanced emissions fromthe coastal surf zone and has treated coarse sea-salt particlesas dry and chemically inert (Sarwar and Bhave 2007)

Studies that have simulated the chemical evolution of sea-salt particles have used alternative models to CMAQ (eg Ja-cobson 1997 Lurmann et al 1997 Meng et al 1998 Sunand Wexler 1998b Sartelet et al 2007 Athanasopoulou etal 2008 Pryor et al 2008) or variants of CMAQ such asCMAQ-MADRID (Zhang et al 2004) These studies of-ten suffered from simple estimates of sea-salt emissions ordid not evaluate model results against measurements of size-segregated PM composition (ie size-composition distribu-tions) Spyridaki et al (2006) did evaluate size-compositiondistributions but did not account for enhanced emissionsof sea salt from the coastal surf zone Kleeman and Cass

(2001) modeled surf-zone emissions but only evaluated size-composition distributions for particles withDaerolt 18 micromA recent example of a CMAQ variant that treats chemicalprocessing of sea salt is CMAQ-UCD (Zhang and Wexler2008) This model was developed for application in theBay Regional Atmospheric Chemistry Experiment (BRACE)(Nolte et al 2008) Although CMAQ-UCD performed wellin that study the model is not suitable for many applicationsbecause its run speed is about 8ndash10 times slower than thestandard version of CMAQ used for regulatory applicationsDespite the numerous modeling efforts described above aneed exists for a computationally-efficient treatment of sea-salt emissions and chemical evolution in a model where re-sults capture the size-composition distributions observed incoastal environments

The BRACE study was conducted to (1) improve estimatesof atmospheric nitrogen deposition to Tampa Bay FL (2) ap-portion nitrogen deposition to local and non-local sourcesand (3) assess the impact of utility controls on nitrogen de-position to Tampa Bay (Atkeson et al 2007) Excessive ni-trogen addition to waterways from the atmosphere and landcan produce eutrophic conditions detrimental for aquatic life(eg low dissolved O2 and high opacity) In 2004 65 ofassessed systems in the continental US had moderate to higheutrophic conditions (Bricker et al 2007) Due to the differ-ent deposition velocities of gases and particles condensationof HNO3 and NH3 on coarse sea salt can alter nitrogen de-position to sensitive ecosystems (Pryor and Sorensen 2000Evans et al 2004) Studies that apportion nitrogen depo-sition to potentially controllable sources could benefit frommodels that accurately and efficiently calculate the chemicalprocessing and deposition of sea salt

Air quality models require good predictions of particlesize distributions to accurately predict dry deposition Ac-curate size distributions are also important to the ongoingdevelopment of an inline photolysis module for CMAQ (Fo-ley et al 2010) and the coupled meteorology and chemistrymodel WRF-CMAQ (Pleim et al 2008) which calculatethe impact of atmospheric particles on radiative transfer andclouds Lung dosimetry models also require information onparticle size because deposition patterns in the lung dependstrongly on particle diameter in addition to flow variables andlung morphometry (Asgharian et al 2001) Due to the reg-ulatory emphasis on mass-based PM concentrations particlesize distributions from the CMAQ model are rarely evaluatedagainst observations In cases where they have been evalu-ated (Elleman and Covert 2009 Park et al 2006 Zhang etal 2006) the focus has been on number or volume distribu-tions of fine particles The availability of size-resolved PMcomposition measurements from the BRACE campaign thatspan two-orders of magnitude (018lt Daerolt 18 microm) pro-vides an opportunity to evaluate CMAQ predictions of size-composition distributions in a coastal urban environment

In this study CMAQ is updated for the version 47 publicrelease to include enhanced emissions of sea-salt particles




2 Modeling

21 Aerosol modeling

























4 anions






SurfminusZone Flux

GongClarkede Leeuw

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=001 ms


10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=90 ms


dV

dD

p (

m3 μ

mminus1

mminus2

sminus1

)











22


22









3 Observations




All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References
































































































2 Modeling

21 Aerosol modeling

























4 anions






SurfminusZone Flux

GongClarkede Leeuw

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=001 ms


10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=90 ms


dV

dD

p (

m3 μ

mminus1

mminus2

sminus1

)











22


22









3 Observations




All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References














































































































4 anions






SurfminusZone Flux

GongClarkede Leeuw

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=001 ms


10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=90 ms


dV

dD

p (

m3 μ

mminus1

mminus2

sminus1

)











22


22









3 Observations




All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References


































































































SurfminusZone Flux

GongClarkede Leeuw

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=001 ms


10minus13

10minus12

10minus11

10minus10

10minus9

RH=080U=90 ms


dV

dD

p (

m3 μ

mminus1

mminus2

sminus1

)











22


22









3 Observations




All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References



































































































22


22









3 Observations




All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References































































































All Sites

v46v46b

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)



4 Results




4 NH+

4 NOminus


















times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References












































































































times100


radic1

nsum








0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References































































































0 5 10 150

5

10

15

SO42minus

v46bv46c

0 1 2 3 40

1

2

3

4

NH4+

0 1 2 3 4 50

1

2

3

4

5

NO3minus

0 1 2 3 40

1

2

3

4

Na+

0 1 2 3 4 5 60

1

2

3

4

5

6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus

0 5 10 15

SO42minus

v46bv46c

0 1 2 3 4

NH4+

0 1 2 3 4 5

NO3minus

0 1 2 3 4

Na+

0 1 2 3 4 5 6

Clminus


Pre

dic

ted

To

tal (

μg m

minus3)





4 NH+

4 NOminus





0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References































































































0005101520253035 SO4


0002040608101214

NH4+


0002040608101214

NO3minus


0002040608101214


05 2 1 5 200002040608101214


SO42minus


NH4+


NO3minus



05 2 1 5 20


SO42minus


NH4+


NO3minus



05 2 1 5 20



dM

(μg

mminus3

)d

lnD

p













01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References





































































































01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17Day (May 1minus17 2002)

00

02

04

06

08



00

02

04

06

08

10



(HNO3+NOminus

3 )] at the Sydney






00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References































































































00

02

04

06

08

10

NO3minus Na+


0002040608101214 Clminus Na+

00

05

10

15 SO42minus Na+

1 2 5 10 18

00

05

10

15

20

25NH4

+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+

NO3minus Na+


Clminus Na+

SO42minus Na+

1 2 5 10 18

NH4+ Na+


Mo

lar

Rat

io




Na+ and SO2minus

4

Na+

ratios in seawater







(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References

































































































(HNO3+NOminus






44 Model timing


5 Closing remarks












Edited by O Boucher

References






































































































Edited by O Boucher

References






























































































Simulating emission and chemical evolution of coarse sea …...J. T. Kelly et al.: Simulating...

Documents

Transcript of Simulating emission and chemical evolution of coarse sea …...J. T. Kelly et al.: Simulating...