The Chemistry CATT-BRAMS model (CCATT-BRAMS 4.5): a ... · BRAMS, this model is conceived to run at...

Geosci Model Dev 6 1389ndash1405 2013wwwgeosci-model-devnet613892013doi105194gmd-6-1389-2013copy Author(s) 2013 CC Attribution 30 License

EGU Journal Logos (RGB)

Advances in Geosciences

Open A

ccess

Natural Hazards and Earth System

Sciences

Open A

ccess

Annales Geophysicae

Open A

ccess

Nonlinear Processes in Geophysics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Discussions

Atmospheric Measurement

Techniques

Open A

ccess

Atmospheric Measurement

Techniques

Open A

ccess

Discussions

Biogeosciences

Open A

ccess

Open A

ccess

BiogeosciencesDiscussions

Climate of the Past

Open A

ccess

Open A

ccess

Climate of the Past

Discussions

Earth System Dynamics

Open A

ccess

Open A

ccess

Earth System Dynamics

Discussions

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

Discussions

GeoscientificModel Development

Open A

ccess

Open A

ccess

GeoscientificModel Development

Discussions

Hydrology and Earth System

Sciences

Open A

ccess

Hydrology and Earth System

Sciences

Open A

ccess

Discussions

Ocean Science

Open A

ccess

Open A

ccess

Ocean ScienceDiscussions

Solid Earth

Open A

ccess

Open A

ccess

Solid EarthDiscussions

The Cryosphere

Open A

ccess

Open A

ccess

The CryosphereDiscussions

Natural Hazards and Earth System

Sciences

Open A

ccess

Discussions

The Chemistry CATT-BRAMS model (CCATT-BRAMS 45)a regional atmospheric model system for integrated air quality andweather forecasting and research

K M Longo1 S R Freitas1 M Pirre 2 V Mar ecal3 L F Rodrigues1 J Panetta4 M F Alonso5 N E Rosario6D S Moreira1 M S Gacita1 J Arteta3 R Fonseca1 R Stockler1 D M Katsurayama1 A Fazenda7 and M Bela8

1Centro de Previsao de Tempo e Estudos Climaticos INPE Cachoeira Paulista Brazil2Laboratoire de Physique et Chimie de lrsquoEnvironnement et de lrsquoEspace CNRS-Universite drsquoOrleans OrleansUMR7328 France3Centre National de Recherches MeteorologiqueGroupe drsquoetude de lrsquoAtmosphere MeteorologiqueMeteo-France and CNRS UMR3589 Toulouse France4Divisao de Ciencias da Computacao Instituto Tecnologico da Aeronautica (ITA) Sao Jose dos Campos Brazil5Faculdade de Meteorologia Universidade Federal de Pelotas (UFPEL) Pelotas Rio Grande do Sul Brazil6Departamento de Ciencias Biologicas Universidade Federal de Sao Paulo (UNIFESP) Diadema Sao Paulo Brazil7Instituto de Ciencia e Tecnologia Universidade Federal de Sao Paulo (UNIFESP) Sao Jose dos Campos Sao Paulo Brazil8Laboratory for Atmospheric and Space Physics University of Colorado Boulder USA

Correspondence toK M Longo (karlalongoinpebr)

Received 19 January 2013 ndash Published in Geosci Model Dev Discuss 21 February 2013Revised 23 June 2013 ndash Accepted 3 July 2013 ndash Published 9 September 2013

Abstract Coupled Chemistry Aerosol-Tracer Transportmodel to the Brazilian developments on the Regional Atmo-spheric Modeling System (CCATT-BRAMS version 45) isan on-line regional chemical transport model designed forlocal and regional studies of atmospheric chemistry from thesurface to the lower stratosphere suitable both for operationaland research purposes It includes gaseousaqueous chem-istry photochemistry scavenging and dry deposition TheCCATT-BRAMS model takes advantage of BRAMS-specificdevelopment for the tropicssubtropics as well as the recentavailability of preprocessing tools for chemical mechanismsand fast codes for photolysis rates BRAMS includes state-of-the-art physical parameterizations and dynamic formula-tions to simulate atmospheric circulations down to the meterThis on-line coupling of meteorology and chemistry allowsthe system to be used for simultaneous weather and chem-ical composition forecasts as well as potential feedback be-tween the two The entire system is made of three preprocess-ing software tools for user-defined chemical mechanismsaerosol and trace gas emissions fields and the interpolation ofinitial and boundary conditions for meteorology and chem-istry In this paper the model description is provided alongwith the evaluations performed by using observational data

obtained from ground-based stations instruments aboard air-crafts and retrieval from space remote sensing The evalua-tion accounts for model applications at different scales frommegacities and the Amazon Basin up to the intercontinentalregion of the Southern Hemisphere

1 Introduction

The type of models most commonly used to study and fore-cast atmospheric chemistry is the Eulerian three-dimensionalchemistry transport model (CTM) Global CTMs are used fortropospheric (eg Wild et al 2004 Bousserez et al 2007)and stratospheric research (eg Lefevre et al 1994 Chip-perfield 2006) in particular for monitoring stratosphericozone depletion Regional CTMs are generally designed tostudy air pollution (eg Zhang et al 2006 Honore et al2007) Some of these models are used for operational pur-poses Transport is calculated within CTMs by using off-linewind fields provided by analyses or forecasts computed bya meteorological general circulation model (GCM) Fairlydetailed chemical mechanisms can be used in CTMs be-cause of the low computing cost of the transport component

Published by Copernicus Publications on behalf of the European Geosciences Union

1390 K M Longo et al The Chemistry CATT-BRAMS model (CCATT-BRAMS 45)

However in complex dynamic situations such as convectiveenvironments the use of off-line dynamics can be a majorweakness (Grell and Baklanov 2011)

Thanks to rapid advances in computing resources overthe past ten years models with fully coupled dynamics andchemistry have been developed as follows climate chemistrymodels (CCMs) for climate studies at the global scale (egZeng and Pyle 2003 Eyring et al 2006) and mesoscale(convection-permitting) meteorology-chemistry models forregional work (local studies) (Grell et al 2000 2005 Wangand Prinn 2000 Mari et al 2000 Zhang et al 2003 Fast etal 2006 Arteta et al 2006 Marecal et al 2006 Barth et al2007a) Both types of coupled models must compromise be-tween the spatial resolution (and domain size for limited areamodels) the simulation length and the degree of complexityfor the chemical mechanism Simulations using fine resolu-tions large domains and detailed chemistry over a long dura-tion for both the aerosol and gas phase are still too computa-tionally demanding CCMs usually use coarse horizontal andvertical resolutions with reasonably detailed chemical mech-anisms to run for long periods of time Limited-area modelsimulations at the regional scale or at the cloud scale gener-ally use a reduced number of chemical species and reactionsbecause of their fine horizontal and vertical resolutions (seeexamples of models used in the intercomparison exercise aspublished by Barth et al 2007b)

Here we present a new modeling tool devoted to localand regional studies of atmospheric chemistry from the sur-face to the lower stratosphere that is designed for both op-erational and research purposes This new model representsan advancement in the limited-area model CATT-BRAMS(Coupled Aerosol-Tracer Transport model to the Braziliandevelopments on the Regional Atmospheric Modeling Sys-tem Freitas et al 2009 Longo et al 2010) which includesa chemical module for both gaseous and aqueous phases Italso takes advantage of the BRAMS-specific developmentfor the tropicssubtropics (Freitas et al 2009) of the recentavailability of pre-processing tools for addressing chemicalmechanisms and of fast codes for photolysis rates As withBRAMS this model is conceived to run at horizontal reso-lutions ranging from a few meters to more than a hundredkilometers depending on the scientific objective The modelcomputation is optimized to be used for operational pur-poses This specificity allows for the use of chemical mech-anisms with a large number of chemical compounds for re-search studies

The full model description including code optimizationdetails is given in Sect 2 Two examples of model applica-tions for both regional and urban scale conditions includingthe description of the simulation setups and model resultsare shown and discussed in Sect 3 These two case stud-ies illustrate the modelrsquos ability to simulate the troposphericchemistry of the tropical region Section 4 is devoted to finalremarks

2 Model system description

The Chemistry CATT-BRAMS (hereafter CCATT-BRAMS)is an Eulerian atmospheric chemistry transport model cou-pled on-line with a limited-area atmospheric model The at-mospheric model is the BRAMS (Brazilian developments onthe Regional Atmospheric Modeling System) which is basedon the Regional Atmospheric Modeling System (RAMSWalko et al 2000) with additional specific developmentsfor tropical and subtropical regions (Freitas et al 2009)The Regional Atmospheric Modeling System is a multipur-pose numerical prediction model designed to simulate at-mospheric circulations spanning from the hemispheric scaledown to large eddy simulations of the planetary boundarylayer RAMS is a non-hydrostatic time-split compressiblemodel (Tripoli and Cotton 1982) It has a set of state-of-the-art physical parameterizations for surface-air exchangesturbulence convection radiation and cloud microphysics Ithas a multiple grid-nesting scheme for simultaneously solv-ing model equations on two-way interacting computationalmeshes of differing spatial resolutions For real case studiesglobal meteorological analyses or forecasts are used to de-fine the initial model state and for forcing boundaries duringsimulation

The radiation scheme is based on a modified version of theCommunity Aerosol and Radiation Model for Atmosphere(CARMA Toon et al 1988) The original CARMA versionsimultaneously considered an aerosol microphysics schemeand two-stream radiative transfer module for both solar andterrestrial spectral regions (Toon et al 1989) The majormodification from the original version refers to the prescrip-tion of aerosol intensive optical properties specifically theextinction efficiency single scattering albedo and asymme-try parameter These parameters are obtained from previousoff-line Mie calculations The prescription of smoke opticalproperties especially for the South American continent de-rives from the use of climatological size distribution and thecomplex refractive index from several measurements sites ofthe AErosol RObotic NETwork (AERONET Holben et al1998) in the southern area of the Amazon Basin These prop-erties are used as input to an off-line Mie code to calculate thespectral optical properties required by CARMA (Procopio etal 2003 Rosario et al 2011 2013)

CATT is an Eulerian transport model coupled to BRAMSand designed to study the transport processes associated withthe emission of tracers and aerosols (Freitas et al 2009Longo et al 2010) Tracer transport is consistently run on-line with the atmospheric state evolution using the BRAMSdynamic and physical parameterizations The tracer mass-mixing ratio which is a prognostic variable includes the ef-fects of sub-grid scale turbulence in the planetary boundarylayer convective transport by shallow and deep moist con-vection in addition to grid scale advection transport The ad-vective transport of scalars can employ the forward upstreamof second order formulation (Tremback et al 1987) or the

Geosci Model Dev 6 1389ndash1405 2013 wwwgeosci-model-devnet613892013

K M Longo et al The Chemistry CATT-BRAMS model (CCATT-BRAMS 45) 1391

monotonic formulation developed by Walcek (2000) and im-plemented by Freitas et al (2012) The general mass continu-ity equation for tracers solved in the CATT-BRAMS modelis as follows

parts

partt=

(parts

partt

)adv

+

(parts

partt

)PBLdiff

+

(parts

partt

)deepconv

+

(parts

partt

)shallowconv

+

(parts

partt

)chem

+ W + R + Q (1)

where s is the grid box mean tracer mixing ratio the termadv represents the 3-D resolved transport (advection by themean wind) and the terms PBL diff deep conv and shal-low conv stand for the sub-grid scale turbulence in the plan-etary boundary layer (PBL) and deep and shallow convec-tion respectively In the original version of CATT-BRAMSas described in Freitas at al (2009) the chem term referssimply to the passive tracersrsquo lifetime theW is the term forwet removal applied only to aerosols andR is the term forthe dry deposition applied to both gases and aerosol parti-cles FinallyQ was the source term that already included theplume rise mechanism associated with vegetation fires (Fre-itas et al 2006 2007 2010) In addition this model versionalready accounts for the interaction of aerosols with solarand long wave radiation and the feedback to the atmosphericmodel in terms of heating rates (Longo et al 2006) How-ever the aerosols are treated as generic particles (with diam-eters smaller than 25 microm) from urban and biomass burningsources without considering the evolution of its compositionand size distribution not accounting for interactions betweenaerosol and gas-phase chemistry

A detailed description and evaluation of the CATT-BRAMS can be found in Freitas et al (2009) Longo et al(2010) and Rosario et al (2013) CATT-BRAMS has beenused operationally by the Brazilian Center for Weather Pre-diction and Climate Studies (CPTEChttpmeioambientecptecinpebr) since 2003 and as a research tool by sev-eral groups around the world (eg Marecal et al 20072010 Longo et al 2009 Landulfo et al 2009 Arteta etal 2009a b Liu et al 2010 Ramos 2006 Gevaerd 2005Herrmann 2004 Gacita 2011 Munchow 2008 Hernandez2010 Alonso 2011) which demonstrates its good perfor-mance

In the new CCATT-BRAMS a gas phase chemical moduleis fully coupled to the meteorologicaltracer transport modelCATT-BRAMS to solve the chem term in Eq (1) as follows

(partρk

partt

)chem

=

(dρk

dt

)= Pk (ρ) minus Lk (ρ) (2)

where ρ = ρ1ρ2 ρN are the number density ofNreagent species considered andPk and Lk correspond tothe net production and loss of speciesk This equation usesthe initial condition ρ(t0) = ρ1(t0)ρ2(t0) ρN (t0) in

35

Figure 1 Sub-grid processes involved in trace gas emissions transport transformation and

deposition as simulated by the Chemistry ndash CATT-BRAMS system

Fig 1 Sub-grid processes involved in trace gas emissions trans-port transformation and deposition as simulated by the ChemistryCATT-BRAMS system

its solution The loss and production terms include photo-chemistry gas phase and aqueous chemistry Figure 1 il-lustrates the main sub-grid scale processes involved in thetrace gasaerosol distribution in the CCATT-BRAMS systemwhich will be described below

The CCATT-BRAMS system was developed by using ad-vanced numerical tools to come up with a flexible multi-purpose model that can be run for both operational fore-casts and research simulations Moreover the model systemis a fully comprehensive code package designed to allowthe user some flexibility in choosing the chemical speciesthe chemical mechanism the emission types and databasesA schematic view of the model system including its pre-processing tools is given in Fig 2 The packed systemcomprises three preprocessor tools for emissions databases(PREPCHEM SRC Freitas et al 2011) chemical mech-anisms and boundary and initial conditions as well as theCCATT-BRAMS code itself

The first step for any simulation is the choice of gaseousspecies and of the gas phase and photochemistry reactionsTo make the use of different chemistry mechanisms morestraightforward within CCATT-BRAMS a modified versionof the preprocessor tool SPACK (Simplified Preprocessor forAtmospheric Chemical Kinetics) as developed by Djouad etal (2002) following Damian et al (1995) was used Themodified SPACK (hereafter called M-SPACK) basically al-lows the passage of a list of species and chemical reac-tions from symbolic notation (text file) to a mathematicalone using ordinary differential equations (ODEs) automat-ically preprocesses chemical species aggregation and createsFortran90 routine files directly compatible for compilationwithin the main CCATT-BRAMS code

The M-SPACK output code also feeds the emissions pre-processor tool (PREP-CHEM-SRC) to ensure consistencybetween the emissions database to be used in CCATT-BRAMS and the list of species treated by the chemical mech-anism The PREP-CHEM-SRC provides the emissions fromdifferent sources (see Sect 22 for details) which are inter-polated onto the chosen simulation grid and then used as

wwwgeosci-model-devnet613892013 Geosci Model Dev 6 1389ndash1405 2013


Fig 2 Schematic of the CCATT-BRAMS system The gray blocksand the black arrows indicate the codes that make up the CCATT-BRAMS System and their outputs respectively The white blocksindicate either the input files for the preprocessing (first line) orthe preprocessing outputs (third line) which are also input files forpreprocessing emissions and boundary conditions and routines forcomposing the CCATT-BRAMS model

inputs for the CCATT-BRAMS This model basically linksthe species that are in the emissions databases to those inthe chemical mechanism When the species are the samethis is straightforward However there are often problems inwhich the species names represent a group of species In thiscase the PREP-CHEM-SRC converts the species from thedatabase into the model species For volatile organic com-pounds (VOCs) when only the total emissions per sourceare reported the disaggregation follows the OxComp (tropo-spheric Oxidant model Comparison Wild and Prather 2000Derwent et al 2001) specification for the hydrocarbon mix-ture as reported in the Third Assessment Report of the IPCC(Prather et al 2001) The resulting disaggregated VOCs aresubsequently reassembled into lumped species via compati-bility files specific for the chosen chemical mechanism

The M-SPACK output code also feeds the BC-PREP codethat generates the initial and boundary fields for the chemicalspecies to be used in the CCATT-BRAMS simulation

21 Gas phase and photochemistry

In principle CCATT-BRAMS can employ any chemicalmechanism provided by the user In practice the avail-able computing resources limit the number of species andreactions To date three widely used tropospheric chem-istry mechanisms have been tested that is the RegionalAtmospheric Chemistry Mechanism (RACM Stockwell etal 1997) with 77 species Carbon Bond (CB Yarwoodet al 2005) with 36 species and the Regional Lumped

Atmospheric Chemical Scheme (RELACS Crassier et al2000) with 37 species The RACM mechanism is preferen-tially used for research purposes because it is more com-plete than Carbon Bond and RELACS Because RACM andRELACS were used in the simulations presented in this pa-per more details will be given on these mechanisms RACMis based on the lumped molecule approach in which thespecies are grouped according to their chemical nature alsousing the magnitude of the emission rates and reactivity withrespect to OH in the lumping from the included speciesThis model was originally designed to simulate tropospherechemistry from urban to remote conditions and it includes17 stable inorganic species 4 inorganic intermediates 32 sta-ble organic species and 24 organic intermediates for a totalof 237 reactions including 23 photolytic reactions RELACSis a reduction of RACM which follows the reactivity lumpedmolecular approach but modified so that the RACM or-ganic species are grouped into new species according to theirchemical nature and the reactivity is relative not only toOH but also to other oxidants The result is a reduction to37 species and 128 reactions RELACS is currently used foroperational air quality prediction at CPTEC because of itsreduced mechanism leading to affordable computing timesFor real-time forecasting RELACS reproduces the RACMresponse fairly well (Gacita 2011)

Most atmospheric chemistry models have used look-up ta-bles of pre-calculated photolysis rates because their on-linedetermination by radiative code is time consuming Never-theless this approach does not account for the presence ofaerosols and clouds which are frequent features in meteo-rological simulations in particular in the tropics The devel-opment of fast radiative codes such as Fast-J (Wild et al2000 Brian and Prather 2002) and Fast-TUV (Tie et al2003) provides the opportunity to couple the photolysis pro-cesses calculations with a meteorological model on-line toimprove the photolysis rates estimations in aerosol pollutedand cloudy environments The Fast-TUV code chosen and in-troduced to the CCATT-BRAMS is based on the well-knownTUV model (Madronich 1989) It uses a reduced number ofspectral bands (17) leading to 8 times more rapid code thanthe original TUV with differences generally lower than 5

22 Emissions

The PREP-CHEM-SRC tool provides emission fields inter-polated onto the transport model grid The preprocessor iscoded using Fortran90 and C and is driven by a namelist al-lowing the user to choose the type of emissions databasesand chemical species specification file from the M-SPACKThe emissions under consideration are from the most recentdatabases of urbanindustrial biogenic biomass burningvolcanic biofuel use and burning from agricultural wastesources The current version includes anthropogenic emis-sion inventories provided by ldquoREanalysis of the TROpo-spheric chemical composition over the past 40 yrrdquo (RETRO



httpretroenesorg) for 26 species and is complementedby other species provided by the ldquoEmission Database forGlobal Atmospheric Research ndash version 42rdquo (EDGAR-42httpedgarjrceceuropa Olivier et al 1996 1999) Thisversion includes most direct greenhouse gases 4 ozone pre-cursor species and 3 acidifying gases primary aerosol par-ticles and stratospheric ozone depleting species It also in-cludes urbanindustrial information specifically for the SouthAmerican continent based on local inventories (Alonso etal 2010) Currently all urbanindustrial emissions are re-leased in the lowest model layer However if emissionsfrom point sources (eg stacks) are available information onstack heights can be easily included For biomass burningthe PREP-CHEM-SRC includes emissions provided by theGlobal Fire Emissions Database (GFEDv2) based on Giglioet al (2006) and van der Werf et al (2006) or emissionscan also be estimated directly from satellite remote sensingfire detections using the Brazilian Biomass Burning Emis-sion Model (3BEM Longo et al 2010) as included in thetool In both cases fire emissions are available for 107 dif-ferent species The biomass burning emission estimate is di-vided into two contributions namely smoldering which re-leases material in the lowest model layer and flaming whichmakes use of an embedded on-line 1-D cloud model in eachcolumn of the 3-D transport model to determine the verti-cal injection layer In this case the cloud model is integratedusing current environmental conditions (temperature watervapor mixing ratio and horizontal wind) provided by the hostmodel (Freitas et al 2006 2007 2010) Biogenic emissionsare also considered via the Global Emissions Inventory Ac-tivity of the Atmospheric Composition Change the Euro-pean Network (GEIAACCENThttpwwwaerojussieufrprojetACCENTdescriptionphp) for 12 species and derivedby the Model of Emissions of Gases and Aerosols from Na-ture (MEGAN Guenther et al 2006) for 15 species Otheremissions include volcanic ashes (Mastin et al 2009) vol-canic degassing (Diehl 2009 Diehl et al 2012) biofueluse and agricultural waste burning inventories developed byYevich and Logan (2003)

The PREPCHEM SRC code is a comprehensive user-friendly tool and a regular user should be able to incorpo-rate new emissions databases A detailed description of thePREPCHEM SRC code and its functionalities can be foundin Freitas et al (2011)

23 Transfer to aqueous phase and aqueous chemistry

In the presence of liquid clouds or rain soluble species arepartly transferred from the gas phase to the liquid phase Inthe microphysical scheme the mass transfer calculations be-tween the phases follow Barth et al (2001) At each timestep it is determined either by Henryrsquos law equilibrium or bymass transfer limitation calculations if equilibrium is not at-tained within the time step The cloud and rain droplets trans-port the species in the liquid phase They can be released to

the gas phase if droplets evaporate or wash out by rain sedi-mentation

In the convection scheme the species can be washed outby rain sedimentation The loss of gas species is proportionalto the gas species concentration to the effective Henryrsquos con-stant of the species to the mixing ratio of liquid cloud andrain and to the rain rates following the work of Berge (1993)

The chemical transformations in the aqueous phase areresolved following the scheme proposed by Strader et al(1999) which include 28 species in the aqueous phase asrelated to 18 species in the gaseous phase as described inSartelet et al (2007) and Mallet et al (2007)

24 Dry deposition

Dry deposition of gas species at the surface is taken into ac-count in the CCATT-BRAMS The deposition flux followsthe resistance formulation and accounts for the aerodynamicquasi-laminar layer and canopy resistances (Wesely 1989Seinfeld and Pandis 1998) with updates from Weseley andHicks (2000) and Zhang et al (2003) Additionally it is fullycoupled to the surface parameterization including the sub-grid land type patches within the parameterization

25 Carbon cycle

The soilvegetation model ldquoJoint UK Land EnvironmentSimulatorrdquo (JULES Best et al 2011 Clark et al 2011)was also fully coupled to the CCATT-BRAMS model pro-viding surface fluxes of momentum latent and sensible heatand radiative as well as CO2 and other trace gases A de-tailed description of the coupling and applications to numer-ical weather forecasting and the CO2 budget in South Amer-ica is given in Moreira et al (2013)

26 Time integration of the chemical mechanism

The system of Eq (2) is a stiff one because the chemicalreactions rates occurring in the atmosphere vary by sev-eral orders of magnitude requiring implicit methods fornon-prohibitive numerical solutions Furthermore the typi-cal chemical mechanisms suitable for the troposphere on aregional scale include a large number of species for exam-ple RACM has 77 species reacting via 230 kinetic and pho-tolysis reactions (Stockwell et al 1997) This implies a highcomputational cost involved in the solution of Eq (2) whichis by far the most expensive term for the solution of the con-tinuity Eq (1)

The numerical integrator of the chemical mechanism inthis numerical modeling system is an efficient implicit andmulti-stage solver based on Rosenbrockrsquos method (Hairerand Wanner 1991 Verwer et al 1999) This method makespossible the change of solution in a nonlinear differentialequation system (Eq 2) to a linear algebraic increment in



terms ofKi and the solution is given as follows

ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)

wheres is the total number of stagesbi is numerical con-stants depending ons andτ is the time step The incrementsKi are obtained sequentially throughout the solution of thelinear algebraic system given by the following

Ki = τF (ρi) + τJ(ρi (t0)) middot

isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)

F (ρi) = P (ρi) minus L(ρi) (6)

wherei = 1 s αij andγij are constants that depend ons ρi

is an intermediate solution used to recalculate the net produc-tion on stagei given by the termF (ρi) andJ is the Jacobianmatrix of the net production at timet0 The majority of thecomputational effort involves finding the solution for this lin-ear system which has the following basic form

A middot x = b (7)

whereA is anN timesN matrix (N is the number of species)x

is the vector solution andb is the vector of the independentterms Fortunately two properties of this system of equationscan be used to speed up the process The first one is the factthat the Jacobian matrix in Eq (4) is extremely sparse typi-cally only approximately 10 of its elements are non-zerosThe second relevant aspect is that the matrix structure is in-variant in time and space which allows the prior recognitionand mapping of non-zero elements on which the operationsshould be performed The solution for the system of Eq (5)follows the method proposed by Kundert (1986) and Kundertand Sangiovanni-Vincentelli (1988) In this method with ref-erence to fixed positions where the chemical reactions occurone performs the LU factorization and solves the matrix sys-tem by working with memory pointers The method workson three matrices originating from the matrix that must besolved that is a diagonal matrix (D) a lower triangular (L )and another upper triangular (U)

A = D + L + U (8)

The solution given by the following expression

x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)

where (k = 12 ) The elements of the array are dynami-cally allocated only once and unnecessary iterations are elim-inated An index of pointers was previously created to relatethe sequence of positions allocated in the memory with itsoriginal position in the array The factorization order of the

rows and columns of a matrix is of utmost importance anddirectly affects the timeliness and accuracy of the result Themethod solves the array in two parts with decomposition (orfactorization) and with the integrator that uses the substitu-tion forwardndashbackward method

Currently ROS 2 (2nd order 2 stages) and RODAS 3(3rd order 4 stages) Rosenbrockrsquos methods are implementedThe time integration may use a manual splitting or dynamictime step for the chemistry The operator splitting used tosolve the mass continuity equation may be defined as paral-lel (as originally used by the BRAMS model to integrate thedynamic) sequential and sequential symmetric (Yanenko1971 McRae et al 1982 Lanser and Verwer 1998) Thesequential splitting solution is known to be more accuratethan parallel and therefore should be preferred However wealso implemented and tested the sequential symmetric split-ting performing the chemistry integration at each 2 or moretime steps of the dynamic using time steps 2 4 and 6 timeshigher than the dynamic time step These tests indicated thatthe numerical solutions using different choices of splittingare only slightly different typically less than 5 even forozone nearby its precursors emission sources After all thebest compromise between accuracy and computing time wasachieved using RODAS 3 and symmetric splitting with achemistry time step 4 times larger than the time step of thedynamic This setting is used operationally at CPTEC

27 Model data structure

The BRAMS original data structure uses Fortran902003-derived data types to store related model fields Each deriveddata type encapsulates the set of history carrying variablesrequired by one model component For example the deriveddata typeradiation contains radiation-specific fields that areused and updated at each invocation of the radiation modulePrognostic fields are stored in a derived data type namedba-sic as referenced by every module that requires prognosticfield values As a general BRAMS design and coding rulederived data types are used by every model component thatrequires their values but is modified only by its own modelcomponent

This data structure simplifies the introduction of newmodel parameterizations by encapsulating all history-carrying variables required by the new parameterization ata derived data type

BRAMS multiple grids are implemented by arrays of thesederived data types and are indexed by grid number As anexample assuming that variablengrid contains the currentgrid number assuming thatbasicg is the array of the de-rived typebasicthat stores the prognostic fields of all gridsand assuming thatu is the component ofbasic that storesthex component of the wind velocity at all grid points thenbasicg(ngrid)ustores thex component of the wind veloc-ity of the current grid andbasicg(ngrid)u(zxy)addressesthe variable at grid point (z x y)



The CCATT-BRAMS data structure extends the originalBRAMS data structure in a natural way Derived data typechemcontains variables related to one chemical species suchas current mass concentration its tendency its decrease inrelation to dry and wet deposition etc Letsc p be oneof these fields Multiple grids are represented by an arrayof chem type However because each species requires itsset of variables at each grid thechemarray has to be in-dexed by species number and grid number Consequentlychemg(nspeciengrid)scp(zxy)stores fieldsc p at gridpoint (z x y) of speciesnspecieat gridngrid

The (z x y) memory organization eases code develop-ment Modelers can develop new code components for a sin-gle atmospheric column (thez direction) dealing with rankone arrays Then a wrapper is written to connect the newcode component to the existing code The wrapper loops overcolumns invoking the code component with one column at atime Unfortunately most code components have vertical de-pendencies preventing vectorization and increasing execu-tion time To reduce execution time by increasing the vector-ization ratio selected components are rewritten to collapsethe x and y dimension in just one dimension (xy) and totranspose the (z x y) organization into (xy z) followingthe work of Fazenda et al (2006) The wrapper transposesthe (z x y) memory organization into the (xy z) and backisolating the execution time optimization

Chemistry is a grid point computation (that is computa-tion at one grid point is fully independent of any other gridpoint) Thus the chemistry module could have been writtenon scalar variables dealing with a single grid point at a timebut in a way that would prevent vectorization To increase thevectorization ratio the chemistry module operates over a setof grid points at a time as represented by vectors of rank oneThe driver breaks the (z x y) field organization into vectorsof a convenient size invokes the module and stores back theresults

28 Initial and boundary conditions

The initial and boundary conditions for the chemical speciesare treated in a similar way as for the meteorologicalvariables The RAMSISAN (ISentropic ANalysis packageTremback 1990) tool originally developed to create initialand boundary conditions for the meteorological fields wasextended to include chemical fields from global chemistrymodels This model allows the boundary condition data tobe applied by varying in time using a Newtonian relaxation(nudging) technique

3 Examples of applications

The performance of the CCATT-BRAMS system is exploredthroughout two case studies from the regional to local scaleand from biogenic and fire emissions to urban emission im-pacts

31 Regional scale Amazonian case study

During the LBABARCA (Large-Scale Biosphere Atmo-sphere Experiment in the Amazon BasinBalanco At-mosferico Regional de Carbono na Amazonia Andreae et al2012 Beck et al 2012 Bela et al 2013) field experimentmeasurements of several trace gases including O3 (ozone)CO (carbon monoxide) CH4 (methane) CO2 (carbon diox-ide) and aerosols were made aboard the INPE ldquoBandeiranterdquoaircraft Flights from Manaus transected the Amazon Basinfrom the tropical Atlantic Ocean shore to the northwesternpart of the Amazon Basin passing over a variety of ecosys-tems from the surface to approximately 45 km in altitude(Fig 3) BARCA Phase A occurred in November 2008 dur-ing the transition between the dry and wet seasons with rain-fall to the west and south of Manaus but much drier condi-tions and widespread vegetation fires in eastern and north-ern Amazonia During this period the Inter-tropical Conver-gence Zone (ITCZ) was north of the studied region whichtherefore received predominantly Southern Hemisphere airmasses with a strong contribution of smoke pollutants fromthe African continent Phase B occurred in May 2009 duringthe opposite seasonal transition from the wet to dry seasonwhen the ITCZ was located over the Equator abnormallysouth of its climatological position favoring events of North-ern Hemisphere inflow to Amazonia The precipitation wasabove average and scattered fires occurred mainly in the farsouth

Here we explore the O3 data measured during BARCAFor the model CO evaluation we refer to Andreae et al(2012) in which BARCA A CO measurements were pre-viously compared with results from several different mod-els namely the Stochastic Time Inverted Lagrangian Trans-port (STILT Lin et al 2003) a combination of the WeatherResearch and Forecasting model with Chemistry (WRF-CHEM Grell et al 2005) and the Green House Gasmodule (GHG Beck et al 2011) the HYbrid Single-Particle Lagrangian Integrated Trajectory program (HYS-PLIT4 Draxler and Rolph 2003) and WRF-CHEM andCCATT-BRAMS with 3BEM fire emission sources (Longoet al 2010) Overall all the limited area models includ-ing CCATT-BRAMS with 3BEM fire emissions captured atleast some of the features of the observed mixing ratios in-cluding the general levels of CO enhancement and regionalvariations The dominant sources were fire activities in theeastern and southern part of the Amazon which were wellcaptured by 3BEM with an estimated accuracy of approxi-mately 20 This model comparison indicated that the useof 3BEM emissions in high resolution models includingCCATT-BRAMS yielded results consistent with observa-tions versus those from GFED with a large underestimationfor GFEDv2 and GFEDv3 emissions with factors of four andseven respectively Vertical transport was a less consistentaspect of the modeling because of its strong dependence onthe ability of the model to simulate convection over tropical



37

Figure 3 Flight tracks of the BARCA A (red) and B (black) aircraft campaigns Fig 3 Flight tracks of the BARCA A (red) and B (black) aircraftcampaigns

regions plume rise and emissions All 6 models under studyfailed equally in accurately simulating the vertical distribu-tion More detailed CCATT-BRAMS model evaluation forbiomass burning CO and aerosols over the Amazon area andcentral part of South America for different time periods havebeen shown by Freitas et al (2009) Longo et al (2010) andRosario et al (2013)

In this work BARCA O3 data were disaggregated fol-lowing a similar procedure to that adopted by Andreae etal (2012) for CO Vertical average profiles within four re-gions were considered E and W of Manaus ndash within 4 ofthe Equator and N and S of Manaus ndash roughly on the 60 Wmeridian

We also compared model results with mean troposphericO3 retrieved from the Aura Ozone Monitoring Instru-ment and the Microwave Limb Sounder measurements(OMIMLS) which are available globally on a 1

times 125

(latitude by longitude) grid size (Chandra et al 2003 2004)For model comparison the tropopause was calculated ac-cording to the WMO (World Meteorological Organization)2 Kkmminus1 lapse rate definition by the National Center for En-vironmental Prediction (NCEP) Climate Diagnostics Center

311 Model configuration

For this regional scale case study the model was run from1 October to 31 November 2008 and 1 May to 30 June 2009with two nested grids The external one had a 140 km reso-lution including both the South American and African conti-nents and the internal one had a 35 km resolution includingonly South America This configuration attempts to improvethe description of long-range pollutant transport from Africato South America by providing better boundary conditionsof chemical fields to the nested grid domain However itmust be said that there is an unavoidable inconsistency be-tween the fire data sets used for emissions estimations in

Africa and South America While in South America thereare multi-sensor fire data available (for details see Longoet al 2010) only the Moderate Resolution Imaging Spec-troradiometer (MODIS) fire count was available for AfricaIn Fig 4 the model grid domains for this case study are pre-sented with the total emissions for carbon monoxide obtainedfrom PREP-CHEM-SRC (Freitas et al 2011) using 3BEM(Longo et al 2010) fire emissions as well as all the othersdescribed in Sect 22

The model parameterizations chosen for the simulationsdescribed in this section are detailed as follows The sur-face scheme was the Land Ecosystem-Atmosphere Feedbackmodel (LEAF v3 Walko et al 2000) tuned for tropical ar-eas The RAMS parameterization for the unresolved turbu-lence in the PBL used in this simulation was based on theMellor and Yamada (1982) formulation which predicts tur-bulent kinetic energy (TKE) All radiative calculations wereperformed using CARMA (Toon et al 1989) with aerosoloptical properties prescribed accordingly to Rosario et al(2013) For the microphysics we used the single-momentbulk microphysics parameterization which includes cloudwater rain pristine ice snow aggregates graupel and hail(Walko et al 1995) It includes prognostic equations for themixing ratios of rain and each ice category of total water andthe concentration of pristine ice Water vapor and cloud liq-uid mixing ratios are diagnosed from the prognostic variablesusing the saturation mixing ratio with respect to liquid wa-ter The shallow and deep cumulus scheme is based on themass-flux approach (Grell and Devenyi 2002) The RACMmechanism was used for the chemical integration

The CPTECINPE global model T213 analysis providedinitial and boundary conditions for the meteorological in-tegration and the MOCAGE (Modelisation de la ChimieAtmospherique Grande Echelle) global chemistry forecast(Josse et al 2004 Bousserez et al 2007) provided bound-ary conditions for chemistry fields Initial soil moisture wastaken from the Gevaerd and Freitas (2006) estimation tech-nique and the soil temperature was initialized assuming avertically homogeneous field defined by the air temperatureclosest to the surface from the atmospheric initial data

312 Results

The monthly mean tropospheric O3comparison for the outerdomain between the model results and Aura OMIMLS re-trievals for both BARCA periods suggest that the model suc-cessfully captured the general pattern of O3 distribution overthe South Atlantic region though it underestimated O3 (seeFig 5) In both cases the reader must disregard noise in theOMIMLS retrieval over higher latitudes (Jing et al 2006Waters et al 2006)

For the BARCA A period in November 2008 the modelresults indicate the predominance of the inflow of south-west Atlantic air masses into the Amazon Basin which canalso be seen in OMIMLS data (Fig 5a and b) This inflow



Fig 4 CO emission rates (10minus5 kgmminus2dayminus1) for BARCA A (a)and BARCA B (b) periods from PREP-CHEM-SRC within themodel grid domains

brings enriched air masses with biomass burning productsfrom South Africa The eastndashwest gradients of the O3 mix-ing ratio inside the Amazon Basin are similar for both modeland OMIMLS data though there is a bias of approximately30 The mean O3 mixing ratio from the model ranged fromapproximately 30 ppbv in the central part of the AmazonBasin to approximately 50 ppbv on the eastern border of thebasin On the other hand OMIMLS data varied between ap-proximately 45 and 65 ppbv However the underestimationover the basin seems to be associated with an underestima-tion of O3 in the air masses inflow from the South Atlantic by20 ppbv The three SHADOZ (Southern Hemisphere ADdi-tional OZonesondes Thompson et al 2007) O3 profiles (notshown) obtained during this period (7 19 and 28 November2008) at the Natal (Brazil) site (54 S 354 W) also indi-cate an inflow of enriched O3 air masses above 700 hPa witha typical vertical gradient going from approximately 25 ppbvbelow this level to 80 ppbv above on average but reaching100 ppbv Although the Natal SHADOZ profiles cannot becompared directly with the BARCA measurements they dogive an indication of the inflow air masses from the SouthAtlantic These O3-rich air masses were not intercepted byBARCA flights which were mainly below 700 hPa (Fig 6)

In Fig 6 the typical O3 values observed below 45 km werecaptured by the model when compared to BARCA measure-ments

The underestimation of O3 in the mid-troposphere overthe South Atlantic region is likely related to an underesti-mation of ozone precursors CO in particular According toAndreae et al (2012) BARCA CO measurement compar-isons with CCATT-BRAMS and the Weather Research andForecasting model coupled with Chemistry (WRF-CHEM)results both using 3BEM fire emissions indicate that fire ac-tivities in the eastern and southern part of the Amazon werewell captured by 3BEM with an estimated accuracy of ap-proximately 20 However a much larger underestimationof 3BEM emissions is expected for Africa compared to SouthAmerica and is associated with the differencesinconsistencybetween the fire remote sensing data sets used Based on thefire data input used by 3BEM for Africa one should expectit to be more comparable with GFED which according An-dreae et al (2012) yields a large underestimation of a factorof four and seven for GFEDv2 and GFEDv3 respectively

For the BARCA B period in May 2009 the air massinflow events from the Northern Hemisphere were mostlycaptured on a case-by-case basis but smoothed out in themonthly mean for both the model and Aura data (Fig 5cand d) During this period ozone was measured primarilyin the northern and eastern regions (not shown) The modelalso typically captured the general vertical structure in bothregions

A more detailed discussion of BARCA CO and O3 mea-surements and comparison with CCATT-BRAMS and WRF-CHEM results is given in Andreae et al (2012) and Bela et al(2013) respectively The two models used the same emissionsets as performed with similar accuracy for this case study

32 Local scale urban conditions

This case study was conceived to test a configuration for aplanned air quality forecasting procedure for the major ur-banized region in Brazil which includes the metropolitan ar-eas of Sao Paulo and Rio de Janeiro as well as the highlyurbanized and industrial areas connecting the two

This configuration was evaluated throughout a comparisonof air quality measurement data from several stations oper-ated by the Environmental Agency of the state of Sao Paulo(CETESB Companhia de Tecnologia de Saneamento do Es-tado de Sao Paulo) which continuously operates 42 groundstations in the state of Sao Paulo several inside the Sao Paulometropolitan area (CETESB 2012) All of them measure O3and NOx (nitrogen oxides) and most of the stations in themetropolitan area of Sao Paulo also measure CO


The model was run with a 3 km horizontal resolution from20 July to 30 August 2011 with an 11-day period from



Fig 5Mean tropospheric O3 (ppbv) for November 2008(a b) and May 2009(c d) from the model(a c)and from OMIMLS(b d)

Fig 6 Vertical profiles (regional medians 1000 m blocks) for O3from CCATT-BRAMS model compared to observations disaggre-gated by the E W N and S regions of Manaus

20ndash31 July used for a spin-up The model domain covers alarge part of Sao Paulo and Rio de Janeiro states and thesouthern part of Minas Gerais It was conceived with two pur-poses that is to include the Rio de Janeiro metropolitan areaand its influence on the air quality of the state of Sao Pauloand to provide air quality forecasting for the majority of theimportant cities in the state of Sao Paulo considering thecompromise in the increased computational requirements asnecessary for operational forecasting In Fig 7 the modeldomain is presented with the urban emissions for carbonmonoxide following Alonso et al (2010) as obtained fromPREP-CHEM-SRC (Freitas et al 2011) The location of themetropolitan areas of Sao Paulo Rio de Janeiro Campinasand Ribeirao Preto are also indicated in the same figure

41

Figure 7 CO emission rates (10-5 kg m-2 day-1) from PREP-CHEM-SRC in the model

domain

Figure 8 Monthly average spatial distribution of CO mixing ratio (ppbv) at 0900 AM local

time from the model

13 13 Satildeo13 Paulo13

13 Campinas13

13 Ribeiratildeo13 Preto13

Sorocaba13 13 13

13 Rio13 de13 Janeiro13

Fig 7CO emission rates (10minus5 kgmminus2dayminus1) fromPREP-CHEM-SRC in the model domain

The same model parameterizations as in the previous casestudy were chosen except for the deep convective parame-terization that was not used and for the RELACS chemicalmechanism instead of RACM For the simulations presentedin this section boundary conditions for both meteorologicaland chemistry fields are provided by CPTECINPE CCATT-BRAMS operational forecast data currently with a 25 kmhorizontal resolution (CPTECINPE 2012) by using the one-way nesting approach

322 Results

The monthly average carbon monoxide mixing ratios at thefirst peak hour (0900 LT) are presented in Fig 8 The modelresults indicate high mixing ratios ranging from 1000 to2000 ppbv in the city of Sao Paulo and moderately highones ranging from 500 to 750 ppbv in Rio de JaneiroCampinas and other major cities in the state of Sao Paulo Asimilar behavior is observed at 1800 LT though less intenseand more concentrated around the cities In Sao Paulo a sec-ond peak is observed at approximately 2100 LT and mixingratios stay elevated even at night Nitrogen oxide mixing ra-tios follow a similar diurnal cycle and are also more elevated



Fig 8 Monthly average spatial distribution of CO mixing ratio(ppbv) at 0900 LT from the model

at these hours monthly averages at 0900 LT are presentedin Fig 9 NOx typically peaked above 150 ppbv in the city ofSao Paulo and approximately 100 ppbv in the smaller townsOzone mixing ratios peaked approximately 1500 LT and themonthly averages for this time are presented in Fig 10

Figure 11 presents monthly average diurnal cycle for car-bon monoxide mixing ratios at two stations namely ParqueDom Pedro II and Pinheiros both in the downtown area ofthe city of Sao Paulo The model tends to underestimate thenocturnal observed values of carbon monoxide mixing ratiosat these Sao Paulo stations This underestimation is relatedto the specified diurnal cycle which is not capable of repro-ducing the emissions cycles of Sao Paulo where the trafficat night is less intense than during the day but still not neg-ligible A known difficulty also evident from the results isthat this cycle is not the same throughout the city and theproportion of lightheavy vehicles and the traffic driving pat-terns are not constant as well modifying the emissions Themodel results tend to overestimate the CO for stations lo-cated inside large vegetated areas where observations typ-ically indicate values lower than the average especially atpeak hours Nevertheless the model represents the diurnalcycle during daytime hours for all 10 stations evaluated in-side the Sao Paulo metropolitan area and adequately esti-mates the order of magnitude in most cases with model val-ues within the standard deviation of the monthly mean

The monthly average nitrogen oxides and ozone mixingratios for the Parque Dom Pedro II stations are presented inFig 12 It should be noted that the model overestimates theNOx mixing ratios at peak hours The reasons for this overes-timation are very similar to those for carbon monoxide Ad-ditionally the emission inventories are most likely not accu-rate enough and their spatial distribution in the city accord-ing to the number of vehicles registered for each region alsohas uncertainties when the locations are considered togetherAdditionally only the emissions for NOx and CO are avail-able (Alonso et al 2010) volatile organic compounds (VOC)emissions were updated from local inventories keeping the

Fig 9 Monthly average spatial distribution of NOx mixing ratio(ppbv) at 0900 LT from the model

Fig 10 Monthly average spatial distribution of O3 mixing ratio(ppbv) at 1500 LT from the model

Fig 11 Monthly average CO mixing ratio (ppbv) from twoCETESB ground stations in Sao Paulo city(a) Parque Dom Pe-dro II and(b) Pinheiros and from the model


1400 K M Longo et al The Chemistry CATT-BRAMS model (CCATT-BRAMS 45)μ

Fig 12Monthly (a) NOx mixing ratio (ppbv) and(b) ozone (ppbv)from a CETESB ground station (Parque Dom Pedro II) in Sao Paulocity and from the model

RETRO ratio As a secondary pollutant ozone higher val-ues are expected far from the sources of its precursors (ni-trogen oxides and non-methane volatile organic compoundsndash NMVOC) For all cities a VOC NOx ratio below 4 wasobserved at peak emissions hour as in the VOC-sensitiveregime where the ozone levels were controlled by the VOClevels and a decrease in NOx mixing ratios resulting in higherozone levels This trend was indeed reproduced by the modelwhich showed higher ozone levels in the countryside aroundSao Paulo corresponding to lower NOx emissions ModeledO3 and NOx mixing ratios are both much closer in magnitudeto observed values for the stations within the countrysideFigure 13 illustrates the O3 and NOx diurnal cycles for Soro-caba This finding is similar to all stations in the countrysideas evaluated including Ribeirao Preto which is a larger citycompared to the others with more the 600 000 inhabitants

4 Conclusions

In this paper we have described the functionalities of the newChemistry-CATT-BRAMS modeling system which has beendeveloped by using advanced numerical tools to create a flex-ible multipurpose model that can be run for both operationalforecasts and research purposes The code package includespreprocessor tools for emissions databases chemical mecha-nisms and boundary and initial conditions and the CCATT-BRAMS code itself The model has been developed by fo-cusing with special attention (but not only) on tropical andsubtropical regions The model performance was explored

μ

Fig 13 Monthly (a) NOx mixing ratio (ppbv) and(b) O3 (ppbv)from a CETESB ground station in a small city in the countryside inthe state of Sao Paulo (Sorocaba) and from the model

throughout two different applications for regional and localscales and different emission source influences

For the regional scale a period when BARCA airbornemeasurements were undertaken over the Amazonia was sim-ulated with CCATT-BRAMS In spite of fire emission un-derestimations for Africa which biased the pollution burdenover the South Atlantic region the model was able to capturethe O3 distribution and transport patterns

For the urban case NOx and O3 mixing ratios are over-and underestimated respectively but in all cases they fellwithin the standard deviations Even the nighttime O3 mixingratios which usually represent a greater challenge for mod-elers were accurately assessed However air quality fore-casting for South American cities is still challenging dueto the lack of reliable urban emission inventories Never-theless the model did reproduce the VOC-sensitive regimewith VOC NOx ratios typically below 4 the hour of peakemissions The model results also shows higher ozone lev-els in the countryside from the state of Sao Paulo corre-sponding to lower NOx emissions The CCATT-BRAMScurrently runs for daily operational air quality forecasts atCPTECINPE supporting its robustness It has already beenused for research to study atmospheric chemistry by Marecalet al (2012) Iacono-Marzano et al (2012) and Krysztofiaket al (2012) These studies were based on other chemistryschemes and then discussed in the present paper by using theCCATT-BRAMS capability to easily change the chemistryscheme

The CCATT-BRAMS models system represents a set ofnumerical tools comprising state-of-the-art emission infor-mation as well as physical and chemical parameterizations



coupled to a reliable dynamic core which can cover scalesfrom meters to hundreds of kilometers Moreover the sys-tem is coded with updated Fortran90 Additionally its datastructure is compact and has good readability which makesthe code easily understandable and lower maintenance TheCCATT-BRAMS code package is available upon request byemail atgmaicptecinpebr

AcknowledgementsK M Longo and S R Freitas acknowledgethe partial support of this work by the CNPq (3099222007-03063402011-9) and FAPESP (201213575-9) This work waspartially supported by the European integrated project SCOUT-O3(GOCE-CT-2004-505390) by the LEFEINSU program in France(projects UTLS-tropicale and Tropopause 2009) and by the Inter-American Institute for Global Change Research (IAI) CNR II 2017which is supported by the US National Science Foundation (GrantGEO-0452325) The computer resources were provided by CINES(Centre Informatique National de lrsquoEnseignement Superieur)project pce 2227 and by the Brazilian Institute for Space Research(INPE) The authors thank CETESB and the LBABARCA teamfor making their data available especially Meinrat O Andreae andPaulo Artaxo for CO and O3 data respectively

Edited by H Garny

References

Alonso M F Longo K Freitas S Fonseca R Marecal VPirre M and Klenner L An urban emission inventory forSouth America and its application in numerical modeling of at-mospheric chemical composition at local and regional scales At-mos Environ 44 5072ndash5083 2010

Alonso M F Previsao de Tempo Quımico para a America doSul Impacto das Emissoes Urbanas nas Escalas Local e Re-gional PhD thesis Instituto Nacional de Pesquisas Espaci-ais Sao Jose dos Campos Brazil 188 pp available athttpurlibnet8JMKD3MGP8W39R4K92 last access 27 August2013 2011 (in Portuguese)

Andreae M O Artaxo P Beck V Bela M Freitas S GerbigC Longo K Munger J W Wiedemann K T and Wofsy SC Carbon monoxide and related trace gases and aerosols overthe Amazon Basin during the wet and dry seasons Atmos ChemPhys 12 6041ndash6065 doi105194acp-12-6041-2012 2012

Hernandez R A A Implementacion del Modelo CCATT-BRAMSsobre la Zona Central de Chile y su Evaluacion MSc Disser-tation Departamento de Geofısica Universidad de Chile avail-able at httpwwwcapturauchileclhandle2250106744 lastaccess 27 August 2013 2010 (in Spanish)

Arteta J Cautenet S Taghavi M and Audiffren N Impact oftwo chemistry mechanisms fully coupled with mesoscale modelon the atmospheric pollutants distribution Atmos Environ 407983ndash8001 2006

Arteta J Marecal V and Riviere E D Regional modellingof tracer transport by tropical convection ndash Part 1 Sensitivityto convection parameterization Atmos Chem Phys 9 7081ndash7100 doi105194acp-9-7081-2009 2009a

Arteta J Marecal V and Riviere E D Regional modellingof tracer transport by tropical convection ndash Part 2 Sensitiv-

ity to model resolutions Atmos Chem Phys 9 7101ndash7114doi105194acp-9-7101-2009 2009b

Barth M C Stuart A L and Skamarock W C Numerical simu-lations of the July 10 1996 Stratospheric-Tropospheric Experi-ment Radiation Aerosols and Ozone (STERAO)-Deep Convec-tion experiment storm Redistribution of soluble tracers J Geo-phys Res 106 12381ndash12400 2001

Barth M C Kim S-W Skamarock W C Stuart A L Pick-ering K E and Ott L E Simulations of the redistributionof formaldehyde formic acid and peroxides in the July 101996 STERAO deep convection strom J Goephys Res 112D13310 doi1010292006JD008046 2007a

Barth M C Kim S-W Wang C Pickering K E Ott L EStenchikov G Leriche M Cautenet S Pinty J-P BartheCh Mari C Helsdon J H Farley R D Fridlind A M Ack-erman A S Spiridonov V and Telenta B Cloud-scale modelintercomparison of chemical constituent transport in deep con-vection Atmos Chem Phys 7 4709ndash4731 doi105194acp-7-4709-2007 2007b

Beck V Koch T Kretschmer R Marshall J Ahmadov R Ger-big C Pillai D and Heimann M The WRF Greenhouse GasModel (WRF-GHG) Technical Report No 25 Max Planck In-stitute for Biogeochemistry Jena Germany 2011

Beck V Chen H Gerbig C Bergamaschi P Bruhwiler LHouweling S Rockmann T Kolle O Steinbach J Koch TSapart C J van der Veen C Frankenberg C Andreae M OArtaxo P Longo K M and Wofsy S C Methane airbornemeasurements and comparison to global models during BARCAJ Geophys Res 117 D15310 doi1010292011JD0173452012

Bela M M Longo K M Freitas S R Moreira D Beck VWofsy S C Gerbig C Wiedemann K Artaxo P and An-dreae M O Ozone Production and Transport over the AmazonBasin During Wet and Dry Season Transitions in preparationAtmos Chem Phys 2013

Berge E Coupling of wet scavenging of sulphur to clouds in anumerical weather prediction model Tellus B 45 1ndash221993

Best M J Pryor M Clark D B Rooney G G Essery R LH Menard C B Edwards J M Hendry M A Porson AGedney N Mercado L M Sitch S Blyth E Boucher OCox P M Grimmond C S B and Harding R J The JointUK Land Environment Simulator (JULES) model description ndashPart 1 Energy and water fluxes Geosci Model Dev 4 677ndash699doi105194gmd-4-677-2011 2011

Bousserez N Attie J-L Peuch V-H Michou M Pfister GEdwards D Avery M Sachse G Browell E and Ferrare EEvaluation of MOCAGE chemistry and transport model duringthe ICARTTITOP experiment J Geophys Res 112 D120S42doi1010292006JD007595 2007

Brian H and Prather M J Fast-J2 Accurate simulation of strato-spheric photolysis in global chemistry models J Atmos Chem41 281ndash296 2002

CETESB Air Quality Information System available athttpwwwcetesbspgovbrarqualidade-do-ar32-qualar last access30 November 2012

Chandra S Ziemke J R and Martin R V Tropospheric ozone attropical and middle latitudes derived from TOMSMLS residualComparison with a global model J Geophys Res 108 4291doi1010292002JD002912 2003



Chandra S Ziemke J R Tie X and Brasseur G Elevatedozone in the troposphere over the Atlantic and Pacific oceansin the Northern Hemisphere Geophys Res Lett 31 L23102doi1010292004GL020821 2004

Chipperfield M P New version of the TOMCATSLIMCAT off-line chemical transport model Intercomparison of stratospherictracer experiments Q J Roy Meteor Soc 132 1179ndash1203doi101256qj0551 2006

Clark D B Mercado L M Sitch S Jones C D Gedney NBest M J Pryor M Rooney G G Essery R L H Blyth EBoucher O Harding R J Huntingford C and Cox P M TheJoint UK Land Environment Simulator (JULES) model descrip-tion ndash Part 2 Carbon fluxes and vegetation dynamics GeosciModel Dev 4 701ndash722 doi105194gmd-4-701-2011 2011

CPTECINPE Air Quality forecasting available athttpmeioambientecptecinpebr last access 30 November 2012

Crassier V Suhre K Tulet P and Rosset R Development ofa reduced chemical scheme for use in mesoscale meteorologicalmodels Atmos Environ 34 2633ndash2644 2000

Damian V Sandu A Damian M Carmichael G R and PotraF A KPP ndash A symbolic preprocessor for chemistry kinetics ndashUserrsquos guide Technical report The University of Iowa IowaCityIA52246 1995

Derwent R G Collins W J Johnson C E and Stevenson DS Transient behaviour of tropospheric ozone precursors in aglobal 3-D CTM and their indirect greenhouse effects ClimaticChange 49 463ndash487 2001

Diehl T A global inventory of volcanic SO2 emissions for hindcastscenarios available athttpwikiseasharvardedugeos-chemindexphpVolcanicSO2emissionsfrom Aerocom last access27 August 2013 2009

Diehl T Heil A Chin M Pan X Streets D Schultz Mand Kinne S Anthropogenic biomass burning and volcanicemissions of black carbon organic carbon and SO2 from 1980to 2010 for hindcast model experiments Atmos Chem PhysDiscuss 12 24895ndash24954 doi105194acpd-12-24895-20122012

Djouad R Sportisse B and Audiffren N Numerical simulationof aqueous-phase atmospheric models use of a non-autonomousRosenbrock method Atmos Environ 36 873ndash879 2002

Draxler R R and Rolph G D HYSPLIT (HYbrid Single Parti-cle Lagrangian Integrated Trajectory) Model access via NOAAARL READY Website NOAA Air Resources Laboratory Sil-ver Spring MD USA available athttpwwwarlnoaagovHYSPLIT infophp last access 27 August 2013 2003

Eyring V Butchart N Waugh D W Akiyoshi H Austin JBekki S Bodeker G E Boville B A Bruhl C Chipper-field M P Cordero E Dameris M Deushi M Fioletov VE Frith S M Garcia R R Gettelman A Giorgetta M AGrewe V Jourdain L Kinnison D E Mancini E ManziniE Marchand M Marsh D R Nagashima T Newman P ANielsen J E Pawson S Pitari G Plummer D A RozanovE Schraner M Shepherd T G Shibata K Stolarski R SStruthers H Tian W and Yoshiki M Assessment of tem-perature trace species and ozone in chemistry-climate modelsimulations of the recent past J Geophys Res 111 D22308doi1010292006JD007327 2006

Fast J D Gustafson Jr W I Easter R C Zaveri R ABarnard J C Chapman E G Grell G A and Peckham S

E Evolution of ozone particulates and aerosol direct radia-tive forcing in the vicinity of Houston using a fully coupledmeteorology-chemistry-aerosol model J Geophys Res 111D21305 doi1010292005JD006721 2006

Fazenda A L Panetta J and Enari E H Towards Pro-duction Code Effective Portability among Vector Machinesand Microprocessor-Based Architectures in 18th Interna-tional Symposium on Computer Architecture and High Perfor-mance Computing 2006 Ouro Preto ndash MG Proceedings of18th International Symposium on Computer Architecture andHigh Performance Computing 17ndash20 October 2006 availableat httpieeexploreieeeorgstampstampjsptp=amparnumber=4032411ampisnumber=4032400 last access 13 February 20132006

Freitas S R Longo K M and Andreae M O Impact of in-cluding the plume rise of vegetation fires in numerical simula-tions of associated atmospheric pollutants Geophys Res Lett33 L17808 doi1010292006GL026608 2006

Freitas S R Longo K M Chatfield R Latham D Silva DiasM A F Andreae M O Prins E Santos J C Gielow R andCarvalho Jr J A Including the sub-grid scale plume rise of veg-etation fires in low resolution atmospheric transport models At-mos Chem Phys 7 3385ndash3398 doi105194acp-7-3385-20072007

Freitas S R Longo K M Silva Dias M A F Chatfield RSilva Dias P Artaxo P Andreae M O Grell G RodriguesL F Fazenda A and Panetta J The Coupled Aerosol andTracer Transport model to the Brazilian developments on the Re-gional Atmospheric Modeling System (CATT-BRAMS) ndash Part 1Model description and evaluation Atmos Chem Phys 9 2843ndash2861 doi105194acp-9-2843-2009 2009

Freitas S R Longo K M Trentmann J and Latham D Tech-nical Note Sensitivity of 1-D smoke plume rise models to theinclusion of environmental wind drag Atmos Chem Phys 10585ndash594 doi105194acp-10-585-2010 2010

Freitas S R Longo K M Alonso M F Pirre M MarecalV Grell G Stockler R Mello R F and Sanchez Gacita MPREP-CHEM-SRC ndash 10 a preprocessor of trace gas and aerosolemission fields for regional and global atmospheric chemistrymodels Geosci Model Dev 4 419ndash433 doi105194gmd-4-419-2011 2011

Freitas S R Rodrigues L F Longo K M and Panetta J Im-pact of a monotonic advection scheme with low numerical dif-fusion on transport modeling of emissions from biomass burn-ing Journal of Advances in Modeling Earth Systems 4 M01001doi1010292011MS000084 2012

Gacita M S Estudos numericos de quımica atmosferica paraa regiao do Caribe e America Central comenfase em Cuba(sidinpebrmtc-m18201102142032-TDI) MSc dissertationInstituto Nacional de Pesquisas Espaciais Sao Jose dos CamposBrazil 148 pp available athttpurlibnet8JMKD3MGP8W396SUFP last access 13 February 2013 2011 (in Portuguese)

Gevaerd R Estudo da redistribuicao 3d de gases Aerossois deQueimadas em Roraima em 1998 MSc dissertation Universi-dade de Sao Paulo Brazil 2005 (in Portuguese)

Gevaerd R and Freitas S R Estimativa operacional da umidadedo solo para iniciacao de modelos de previsao numerica da at-mosfera Parte I Descricao da metodologia e validacao RevistaBrasileira de Meteorologia 21 59ndash73 2006 (in Portuguese)



Giglio L van der Werf G R Randerson J T Collatz GJ and Kasibhatla P Global estimation of burned area usingMODIS active fire observations Atmos Chem Phys 6 957ndash974 doi105194acp-6-957-2006 2006

Grell G and Baklanov A Coupled Modeling for ForecastingWeather and Air Quality Atmos Environ 45 6845ndash6851doi101016jatmosenv201101017 2011

Grell G A and Devenyi D A generalized approach to parame-terizing convection combining ensemble and data assimilationGeophys Res Lett 29 1693 doi1010292002GL0153112002

Grell G A Emeis S Stockwell W Schoenemeyer T ForkelR Michalakes J Knoche R and Seidl W Application ofa multiscale coupled MM5chemistry model to the complexterrain of the VOTALP valley campaign Atmos Environ 341435ndash1453 2000

Grell G A Peckham S Schmitz R McKeen S A Frost GSkamarock W and Eder B Fully coupled ldquoonlinerdquo chem-istry within the WRF model Atmos Environ 39 6957ndash6975doi101016jatmosenv200504027 2005

Guenther A Karl T Harley P Wiedinmyer C Palmer P Iand Geron C Estimates of global terrestrial isoprene emissionsusing MEGAN (Model of Emissions of Gases and Aerosols fromNature) Atmos Chem Phys 6 3181ndash3210 doi105194acp-6-3181-2006 2006

Hairer E and Wanner G Solving Ordinary Differential EquationsII Stiff and Differential-Algebraic Problems Springer-VerlagBerlin 1991

Herrmann V Balanco de CO2 na atmosfera da bacia Amazonica opapel dos sistemas convectivos MSc dissertation Universidadede Sao Paulo Brazil 2004 (in Portuguese)

Holben B N Eck T F Slutsker I Tanreı D Buis J PSetzer A Vermote E Reagan J A Kaufman Y J Naka-jima T Lavenu F Jankowiak I and Smirnov A AERONETndash A Federated Instrument Network and Data Archive forAerosol Characterization Remote Sens Environ 66 1ndash16doi101016s0034-4257(98)00031-5 1998

Honore C Rouil L Vautard R Beekmann M BessagnetB Dufour A Elichegaray C Flaud J-M Malherbe LMeleux F Menut L Martin D Peuch A Peuch V-Hand Poisson N Predictability of European air quality Theassessment of three years of operational forecasts and analy-ses by the PREVrsquoAIR system J Geophys Res 113 D04301doi1010292007JD008761 2007

Iacono-Marziano G Marecal V Pirre M Gaillard FArteta J Scaillet B and Arndt N T Gas emissionsdue to magmandashsediment interactions during flood magma-tism at the Siberian Traps Gas dispersion and environmen-tal consequences Earth Planet Sc Lett 357ndash358 308ndash318doi101016jepsl201209051 2012

Jing P Cunnold D Choi Y and Wang Y Summertime tropo-spheric ozone columns from Aura OMIMLS measurements ver-sus regional model results over the United States Geophys ResLett 33 L17817 doi1010292006GL026473 2006

Josse B Simon P and Peuch V H Radon global simulationswith the multiscale chemistry and transport model MOCAGETellus B 56 339ndash356 2004

Krysztofiak G Catoire V Poulet G Marecal V PirreM Louis F Canneaux S and Josse B Detailed mod-

eling of the atmospheric degradation mechanism of very-short lived brominated species Atmos Environ 59 514ndash532doi101016jatmosenv201205026 2012

Kundert K S Sparse matrix techniques and their applications tocircuit simulation in Circuit Analysis Simulation and Designedited by Ruehli A E North-Holland New York 281ndash3241986

Kundert K S and Sangiovanni-Vincentelli A In Sparse userrsquosguide ndash A sparse linear equation Solver Version 13a Berkeleyhttpwwwnetliborgsparsespdoc last access 18 March 20081988

Landulfo E Freitas S R Longo K M Uehara S T andSawamura P A comparison study of regional atmospheric sim-ulations with an elastic backscattering Lidar and sunphotom-etry in an urban area Atmos Chem Phys 9 6767ndash6774doi105194acp-9-6767-2009 2009

Lanser D and Verwer J G Analysis of operator splitting foradvection-diffusion-reaction problems from air pollution mod-eling CWI Technical Report MAS-R 9805 CWI Amsterdam1998

Lefevre F Brasseur G P Folkins I Smith A K and Si-mon P Chemistry of the 1991ndash1992 stratospheric winter three-dimensional model simulations J Geophys Res 99 8183ndash8195 1994

Lin J C Gerbig C Wofsy S C Andrews A E Daube BC Davis K J and Grainger C A A near-field tool for sim-ulating the upstream influence of atmospheric observations theStochastic Time-Inverted Lagrangian Transport (STILT) modelJ Geophys Res 108 4493 doi1010292002jd003161 2003

Liu X M Rivi ere E D Marecal V Durry G Hamdouni AArteta J and Khaykin S Stratospheric water vapour budgetand convection overshooting the tropopause modelling studyfrom SCOUT-AMMA Atmos Chem Phys 10 8267ndash8286doi105194acp-10-8267-2010 2010

Longo K M Freitas S R Dias M A S and Dias P L S Nu-merical modelling of the biomass-burning aerosol direct radia-tive effects on the thermodynamics structure of the atmosphereand convective precipitation in Proceedings of 8 InternationalConference on Southern Hemisphere Meteorology and Oceanog-raphy (ICSHMO) 24ndash28 April Foz do Iguacu Brazil 283ndash2892006

Longo K M Freitas S R Andreae M O Yokelson R and Ar-taxo P Biomass burning long-range transport of products andregional and remote impacts in Amazonia and Global Changeedited by Keller M Bustamante M Gash J Silva Dias PAmerican Geophysical Union Washington DC 186 207ndash2322009

Longo K M Freitas S R Andreae M O Setzer A PrinsE and Artaxo P The Coupled Aerosol and Tracer Transportmodel to the Brazilian developments on the Regional Atmo-spheric Modeling System (CATT-BRAMS) ndash Part 2 Model sen-sitivity to the biomass burning inventories Atmos Chem Phys10 5785ndash5795 doi105194acp-10-5785-2010 2010

McRae G J Goodin W R and Seinfeld J H Numerical so-lution of the atmospheric diffusion equation for chemically re-acting flows J Comput Phys 45 1ndash42 doi1010160021-9991(82)90101-2 1982

Madronich S Photodissociation in the atmosphere 1 Actinic fluxand the effect of ground reflections and clouds J Geophys Res



92 9740ndash9752 doi101029JD092iD08p09740 1989Mallet V Quelo D Sportisse B Ahmed de Biasi M Debry

E Korsakissok I Wu L Roustan Y Sartelet K TombetteM and Foudhil H Technical Note The air quality model-ing system Polyphemus Atmos Chem Phys 7 5479ndash5487doi105194acp-7-5479-2007 2007

Marecal V Riviere E D Held G Cautenet S and FreitasS Modelling study of the impact of deep convection on theutls air composition ndash Part I Analysis of ozone precursors At-mos Chem Phys 6 1567ndash1584 doi105194acp-6-1567-20062006

Marecal V Durry G Longo K Freitas S Riviere E D andPirre M Mesoscale modelling of water vapour in the tropi-cal UTLS two case studies from the HIBISCUS campaign At-mos Chem Phys 7 1471ndash1489 doi105194acp-7-1471-20072007

Marecal V Pirre M Riviere E D Pouvesle N Crowley JN Freitas S R and Longo K M Modelling the reversibleuptake of chemical species in the gas phase by ice particlesformed in a convective cloud Atmos Chem Phys 10 4977ndash5000 doi105194acp-10-4977-2010 2010

Marecal V Pirre M Krysztofiak G Hamer P D and JosseB What do we learn about bromoform transport and chemistryin deep convection from fine scale modelling Atmos ChemPhys 12 6073ndash6093 doi105194acp-12-6073-2012 2012

Mari C Jacob D J and Bechtold P Transport and scavengingof soluble gases in a deep convective cloud J Geophys Res105 22255ndash22268 doi1010292000JD900211 2000

Mastin L Guffanti M Servranckx R Webley P Barsotti SDean K Durant A Ewert J Neri A and Rose W A multi-disciplinary effort to assign realistic source parameters to modelsof volcanic ash-cloud transport and dispersion during eruptionsJ Volcanol Geoth Res 186 10ndash21 2009

Mellor G L and Yamada T Development of a turbulence closuremodel for geophysical fluid problems Rev Geophys 20 851ndash875 doidoi101029RG020i004p00851 1982

Moreira D S Freitas S R Bonatti J P Mercado L MRosario N M E Longo K M Miller J B Gloor M andGatti L V Coupling between the JULES land-surface schemeand the CCATT-BRAMS atmospheric chemistry model (JULES-CCATT-BRAMS10) applications to numerical weather fore-casting and the CO2 budget in South America Geosci ModelDev 6 1243ndash1259 doi105194gmd-6-1243-2013 2013

Munchow G B Impacto da assimilacao de dados de aerossoisno modelo ambiental CATT-BRAMS um estudo de casoda campanha CLAIM MSc dissertation sidinpebrmtc-m19201102031246-TDI 97 pp Instituto Nacional dePesquisas Espaciais Sao Jose dos Campos available athttpurlibnet8JMKD3MGP7W39549E8 last access 13 February2013 18 February 2011 (in Portuguese)

Olivier J Bouwman A van der Maas C Berdowski J VeldtC Bloos J Visschedijk A Zandveld P and Haverlag JDescription of EDGAR Version 20 A set of global emis-sion inventories of greenhouse gases and ozone-depleting sub-stances for all anthropogenic and most natural sources on aper country basis and on a 1 degree times1 degree grid RIVM Re-port 771060002TNO-MEP Report R96119 National Instituteof Public Health and the Environment Bilthoven the Nether-lands 1996

Olivier J Bouwman A Berdowski J Veldt C Bloos J Viss-chedijk A van der Maas C and Zandveld P Sectoral emis-sion inventories of greenhouse gases for 1990 on a per countrybasis as well as on 1 times 1 Environ Sci Policy 2 241ndash2641999

Prather M Ehhalt D Dentener F Derwent R G DlugokenckyE Holland E Isaksen I S A Katima J Kirchhoff V Mat-son P Midgley P M and Wang M Atmospheric Chemistryand Greenhouse Gases Chapter 4 in Climate Change 2001 TheScientific Basis edited by Houghton J T Ding Y GriggsD J Noguer M van der Linden P J Dai X Maskell Kand Johnson C A Cambridge University Press Cambridge UK239ndash287 2001

Procopio A S Remer L A Artaxo P Kaufman Y Jand Holben B N Modeled spectral optical properties forsmoke aerosols in Amazonia Geophys Res Lett 30 2265doi1010292003gl018063 2003

Ramos A M Modelacao Numerica do Transporte de PoluentesAtmosfericos em Portugal e suas Relacoes com as CondicoesMeteorologicas PhD thesis Departamento de Fısica Univer-sidade deEvora Portugal 223 pp 2006 (in Portuguese)

Rosario N E Longo K M Freitas S R Yamasoe M Aand Fonseca R M Modeling the South American regionalsmoke plume aerosol optical depth variability and surface short-wave flux perturbation Atmos Chem Phys 13 2923mdash2938doi105194acp-13-2923-2013 2013

Rosario N M E Variability of aerosol optical properties overSouth America and the impacts of direct radiative effect ofaerosols from biomass burning PhD thesis Institute of Astron-omy Geophysics and Atmospheric Sciences University of SaoPaulo Sao Paulo 2011 (in Portuguese)

Sartelet K N Debry E Fahey K and Roustan YSimulation of aerosols and gas-phase species over Eu-rope with POLYPHEMUS system Part I ndash Model todata comparison for 2001 Atmos Environ 41 6116ndash6131doi101016jatmosenv200704024 2007

Seinfeld J and Pandis S Atmospheric Chemistry and PhysicsJohn Wiley amp Sons Inc New York 1998

Stockwell W R Kirchner F and Kuhn M A new mechanismfor regional chemistry modeling J Geophys Res 102 25847ndash25879 1997

Strader R Lurmann F and Pandis S N Evaluation of secondaryorganic aerosol formation in winter Atmos Environ 39 4849ndash4864 1999

Thompson A M Witte J C Smit H G J Oltmans S J John-son B J Kirchhoff V W J H and Schmidlin F J SouthernHemisphere Additional Ozonesondes (SHADOZ) 1998ndash2004tropical ozone climatology 3 Instrumentation station-to-stationvariability and evaluation with simulated flight profiles J Geo-phys Res 112 D03304 doi1010292005JD007042 2007

Tie X Madronich S Walters S Zhang R Rasch P andCollins W Effects of clouds on photolysis and oxydants in thetroposphere J Geophys Res 108 1ndash25 2003

Toon O B Turco R P Westphal D Malone R and Liu M AMultidimensional Model for Aerosols Description of Computa-tional Analogs J Atmos Sci 45 2123ndash2144 doi011751520-0469(1988)045lt2123ammfadgt20CO2 1988

Toon O B McKay C P Ackerman T P and San-thanam K Rapid Calculation of Radiative Heating Rates



and Photodissociation Rates in Inhomogeneous Multiple Scat-tering Atmospheres J Geophys Res 94 16287ndash16301doi101029JD094iD13p16287 1989

Tremback C Powell J Cotton W and Pielke R The forwardin time upstream advection scheme Extension to higher ordersMon Weather Rev 115 540ndash555 1987

Tremback C J Numerical simulation of a mesoscale convectivecomplex model development and numerical results PhD dis-sertation Dept of Atmospheric Science Colorado State Univer-sity Fort Collins CO 80523 Sci Paper No 465 1990

Tripoli G and Cotton W The Colorado State University three-dimensional cloud-mesoscale model Part I General theoreticalframework and sensitivity experiments J Res Atmos 16 185ndash219 1982

van der Werf G R Randerson J T Giglio L Collatz G JKasibhatla P S and Arellano Jr A F Interannual variabil-ity in global biomass burning emissions from 1997 to 2004 At-mos Chem Phys 6 3423ndash3441 doi105194acp-6-3423-20062006

Verwer J Spee E Blom J Hundsdorfer W A Second-Order Rosenbrock Method Applied to Photochemical Dis-persion Problems SIAM J Sci Comput 20 1456ndash1480doi101137S1064827597326651 1999

Walcek C J Minor flux adjustment near mixing ratio ex-tremes for simplified yet highly accurate monotonic calcula-tion of tracer advection J Geophys Res 105 9335ndash9348doi1010291999JD901142 2000

Walko R L Cotton W R Meyers M P and Harring-ton J Y New RAMS cloud microphysics parameterizationPart I the single-moment scheme Atmos Res 38 29ndash62doi1010160169-8095(94)00087-T 1995

Walko R L Band L Baron J Kittel F Lammers R Lee TOjima D Pielke R Taylor C Tague C Tremback C andVidale P Coupled atmosphere-biophysics-hydrology modelsfor environmental modeling J Appl Meteorol 39 931ndash944doi1011751520-0450(2000)039lt0931CABHMFgt20CO22000

Wang C and Prinn R On the roles of deep convective cloudsin tropospheric chemistry J Geophys Res 105 22269ndash222982000

Waters J W Froidevaux L Harwood R S Jarnot R F PickettH M Read W G Siegel P H Cofield R E Filipiak M JFlower D A Holden J R Lau G K Livesey N J ManneyG L Pumphrey H C Santee M L Wu D L Cuddy D TLay R R Loo M S Perun V S Schwartz M J Stek P CThurstans R P Boyles M A Chandra K M Chavez M CChen G-S Chudasama B V Dodge R Fuller R A GirardM A Jiang J A Jiang Y Knosp B W LaBelle R C LamJ C Lee K A Miller D Oswald J E Patel N C PukalaD M Quintero O Scaff D M Van Snyder W Tope M CWagner P A and Walch M J Early validation analyses ofatmospheric profiles from EOS MLS on the Aura satellite IEEET Geosci Remote 44 1106ndash1121 2006

Wesely M L Parameterizations of surface resistance to gaseousdry deposition in re-gional scale numerical models Atmos Env-iron 23 1293ndash1304 1989

Wesely M L and Hicks B B A A review of the current status ofknowledge on dry deposition Atmos Environ 34 2261ndash22822000

Wild O and Prather M J Excitation of the primary troposphericchemical mode in a global CTM J Geophys Res 105 24647ndash24660 2000

Wild O Zhu X and Prather M J Fast-J accurate simulation ofin and below cloud photolysis in tropospheric chemical modelsJ Atmos Chem 37 245ndash282 2000

Wild O Prather M Akimoto H Sundet J Isaksen I Craw-ford J Davis D Avery M Kondo Y Sachse G andSandholm S Chemical transport model ozone simulations forspring 2001 over the western Pacific regional ozone produc-tion and its global impacts J Geophys Res 109 D15S02doi1010292003JD004041 2004

Yanenko N A The method of fractional steps Springer-VerlagBerlin 1971

Yarwood G Rao S Yocke M and Whitten G Z Updatesto the Carbon Bond chemical mechanism CB05 Final Reportto the US EPA RT-0400675 Novato CA available athttpwwwcamxcompublpdfscb05finalreport120805aspx lastaccess 13 February 2013 2005

Yevich R and Logan J An assessment of biofuel use and burn-ing of agricultural waste in the developing world Global Bio-geochem Cy 17 1ndash21 1095 doi1010292002GB0019522003

Zeng G and Pyle J Changes in tropospheric ozone between 2000and 2100 modeled in a chemistry-climate model Geophys ResLett 30 1392 doi1010292002GL016708 2003

Zhang X Heldson Jr J H and Farley R D Numerical modelingof lightning-produced NOx using an explicit lightning scheme 2Three-dimensional simulation and expanded chemistry J Geo-phys Res 108 4580 doi1010292002JD003225 2003

Zhang Y Liu P Queen A Misenis C Pun B Seigneur C andWu S-Y A comprehensive performance evaluation of MM5-CMAQ for the Summer 1999 Southern Oxidants Study episode ndashPart II Gas and aerosol predictions Atmos Environ 40 4839ndash4855 doi101016jatmosenv200512048 2006



However in complex dynamic situations such as convectiveenvironments the use of off-line dynamics can be a majorweakness (Grell and Baklanov 2011)

Thanks to rapid advances in computing resources overthe past ten years models with fully coupled dynamics andchemistry have been developed as follows climate chemistrymodels (CCMs) for climate studies at the global scale (egZeng and Pyle 2003 Eyring et al 2006) and mesoscale(convection-permitting) meteorology-chemistry models forregional work (local studies) (Grell et al 2000 2005 Wangand Prinn 2000 Mari et al 2000 Zhang et al 2003 Fast etal 2006 Arteta et al 2006 Marecal et al 2006 Barth et al2007a) Both types of coupled models must compromise be-tween the spatial resolution (and domain size for limited areamodels) the simulation length and the degree of complexityfor the chemical mechanism Simulations using fine resolu-tions large domains and detailed chemistry over a long dura-tion for both the aerosol and gas phase are still too computa-tionally demanding CCMs usually use coarse horizontal andvertical resolutions with reasonably detailed chemical mech-anisms to run for long periods of time Limited-area modelsimulations at the regional scale or at the cloud scale gener-ally use a reduced number of chemical species and reactionsbecause of their fine horizontal and vertical resolutions (seeexamples of models used in the intercomparison exercise aspublished by Barth et al 2007b)

Here we present a new modeling tool devoted to localand regional studies of atmospheric chemistry from the sur-face to the lower stratosphere that is designed for both op-erational and research purposes This new model representsan advancement in the limited-area model CATT-BRAMS(Coupled Aerosol-Tracer Transport model to the Braziliandevelopments on the Regional Atmospheric Modeling Sys-tem Freitas et al 2009 Longo et al 2010) which includesa chemical module for both gaseous and aqueous phases Italso takes advantage of the BRAMS-specific developmentfor the tropicssubtropics (Freitas et al 2009) of the recentavailability of pre-processing tools for addressing chemicalmechanisms and of fast codes for photolysis rates As withBRAMS this model is conceived to run at horizontal reso-lutions ranging from a few meters to more than a hundredkilometers depending on the scientific objective The modelcomputation is optimized to be used for operational pur-poses This specificity allows for the use of chemical mech-anisms with a large number of chemical compounds for re-search studies

The full model description including code optimizationdetails is given in Sect 2 Two examples of model applica-tions for both regional and urban scale conditions includingthe description of the simulation setups and model resultsare shown and discussed in Sect 3 These two case stud-ies illustrate the modelrsquos ability to simulate the troposphericchemistry of the tropical region Section 4 is devoted to finalremarks

2 Model system description

The Chemistry CATT-BRAMS (hereafter CCATT-BRAMS)is an Eulerian atmospheric chemistry transport model cou-pled on-line with a limited-area atmospheric model The at-mospheric model is the BRAMS (Brazilian developments onthe Regional Atmospheric Modeling System) which is basedon the Regional Atmospheric Modeling System (RAMSWalko et al 2000) with additional specific developmentsfor tropical and subtropical regions (Freitas et al 2009)The Regional Atmospheric Modeling System is a multipur-pose numerical prediction model designed to simulate at-mospheric circulations spanning from the hemispheric scaledown to large eddy simulations of the planetary boundarylayer RAMS is a non-hydrostatic time-split compressiblemodel (Tripoli and Cotton 1982) It has a set of state-of-the-art physical parameterizations for surface-air exchangesturbulence convection radiation and cloud microphysics Ithas a multiple grid-nesting scheme for simultaneously solv-ing model equations on two-way interacting computationalmeshes of differing spatial resolutions For real case studiesglobal meteorological analyses or forecasts are used to de-fine the initial model state and for forcing boundaries duringsimulation

The radiation scheme is based on a modified version of theCommunity Aerosol and Radiation Model for Atmosphere(CARMA Toon et al 1988) The original CARMA versionsimultaneously considered an aerosol microphysics schemeand two-stream radiative transfer module for both solar andterrestrial spectral regions (Toon et al 1989) The majormodification from the original version refers to the prescrip-tion of aerosol intensive optical properties specifically theextinction efficiency single scattering albedo and asymme-try parameter These parameters are obtained from previousoff-line Mie calculations The prescription of smoke opticalproperties especially for the South American continent de-rives from the use of climatological size distribution and thecomplex refractive index from several measurements sites ofthe AErosol RObotic NETwork (AERONET Holben et al1998) in the southern area of the Amazon Basin These prop-erties are used as input to an off-line Mie code to calculate thespectral optical properties required by CARMA (Procopio etal 2003 Rosario et al 2011 2013)

CATT is an Eulerian transport model coupled to BRAMSand designed to study the transport processes associated withthe emission of tracers and aerosols (Freitas et al 2009Longo et al 2010) Tracer transport is consistently run on-line with the atmospheric state evolution using the BRAMSdynamic and physical parameterizations The tracer mass-mixing ratio which is a prognostic variable includes the ef-fects of sub-grid scale turbulence in the planetary boundarylayer convective transport by shallow and deep moist con-vection in addition to grid scale advection transport The ad-vective transport of scalars can employ the forward upstreamof second order formulation (Tremback et al 1987) or the




parts

partt=

(parts

partt

)adv

+

(parts

partt

)PBLdiff

+

(parts

partt

)deepconv

+

(parts

partt

)shallowconv

+

(parts

partt

)chem

+ W + R + Q (1)




(partρk

partt

)chem

=

(dρk

dt



35

















22 Emissions











24 Dry deposition


25 Carbon cycle








ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)



isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)




A middot x = b (7)



A = D + L + U (8)


x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)






















37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References



































































































































parts

partt=

(parts

partt

)adv

+

(parts

partt

)PBLdiff

+

(parts

partt

)deepconv

+

(parts

partt

)shallowconv

+

(parts

partt

)chem

+ W + R + Q (1)




(partρk

partt

)chem

=

(dρk

dt



35

















22 Emissions











24 Dry deposition


25 Carbon cycle








ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)



isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)




A middot x = b (7)



A = D + L + U (8)


x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)






















37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References









































































































































22 Emissions











24 Dry deposition


25 Carbon cycle








ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)



isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)




A middot x = b (7)



A = D + L + U (8)


x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)






















37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References









































































































































24 Dry deposition


25 Carbon cycle








ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)



isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)




A middot x = b (7)



A = D + L + U (8)


x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)






















37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References



































































































































ρ (t0 + τ) = ρ (t0) +

ssumi=1

biKi (3)



isumj=1

γijKj (4)

ρi = ρ (t0) +

iminus1sumj=1

αijKj (5)




A middot x = b (7)



A = D + L + U (8)


x(k+1)= Dminus1

[minus(L + U)x(k)

+ b] (9)






















37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References














































































































































37





times 125







312 Results





















41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References


















































































































































41


domain


time from the model


13 Campinas13


Sorocaba13 13 13




322 Results















4 Conclusions


μ










Edited by H Garny

References













































































































































4 Conclusions


μ










Edited by H Garny

References




































































































































Edited by H Garny

References

































































































































The Chemistry CATT-BRAMS model (CCATT-BRAMS 4.5): a ... · BRAMS, this model is conceived to run at...

Documents

Transcript of The Chemistry CATT-BRAMS model (CCATT-BRAMS 4.5): a ... · BRAMS, this model is conceived to run at...