The Coupled Aerosol and Tracer Transport The Coupled Aerosol and Tracer Transport model to the Brazilian developments on the Regional model to the Brazilian developments on the Regional

Atmospheric Modeling System Atmospheric Modeling System

(CATT(CATT--BRAMS):BRAMS):

S. Freitas, K. LongoS. Freitas, K. Longo

http://http://www.cptec.inpe.br/meio_ambientewww.cptec.inpe.br/meio_ambiente

Center for Weather Forecasting and Climate Studies Center for Weather Forecasting and Climate Studies -- INPE INPE –– BrazilBrazil

Some subSome sub--grid process involved at gases/aerosols grid process involved at gases/aerosols transport and simulated by CATTtransport and simulated by CATT--BRAMSBRAMS

Eulerian Transport Model :Eulerian Transport Model :CATTCATT--BRAMS Atmospheric ModelBRAMS Atmospheric Model

• in-line Eulerian transport model fully coupled to the atmospheric dynamics• suitable for feedbacks studies• tracer mixing ratio tendency equation

where:

•• adv adv gridgrid--scale advectionscale advection•• PBL PBL turbturb subsub--grid transport in the PBLgrid transport in the PBL•• deep deep convconv subsub--grid transport associated to the deep convection grid transport associated to the deep convection

including downdraft at cloud scaleincluding downdraft at cloud scale•• shallow shallow convconv subsub--grid transport associated to the shallow convectiongrid transport associated to the shallow convection•• W W convective wet removal convective wet removal •• RR sink term associated with dry deposition or csink term associated with dry deposition or chemical hemical

transformationtransformation•• Q Q source emission with plume rise subsource emission with plume rise sub--grid transport.grid transport.

2.5,

PM plumePBL deep shallow rise

advturb conv conv

s s s s sW R Q

t t t t t

∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂ = + + + + + += + + + + + += + + + + + += + + + + + +

∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂∂ ∂ ∂ ∂ ∂

Freitas et al., 1999, 2005, 2006

Grid Scale Advection TermGrid Scale Advection Term

Advection term:

i

iadv i

s su

t x

∂ ∂∂ ∂∂ ∂∂ ∂ = −= −= −= − ∂ ∂∂ ∂∂ ∂∂ ∂ ∑∑∑∑

(((( )))) (((( ))))(((( )))){{{{

(((( )))) (((( ))))(((( ))))}}}}

0 01/ 2 1/ 20

0 01/ 2 1/ 2

1j j

j

j j

su F F

x x

s u u

ρ ρρ ρρ ρρ ρρρρρ

ρ ρρ ρρ ρρ ρ

+ −+ −+ −+ −

+ −+ −+ −+ −

∂∂∂∂ − ≈ − −− ≈ − −− ≈ − −− ≈ − −

∂ ∆∂ ∆∂ ∆∂ ∆

− −− −− −− −

(((( )))) (((( ))))2

1 / 2 1 1

1 / 2

2 2

:

j j j j j

j

xF s s s s

t

tand u

x

α αα αα αα α

αααα

+ + ++ + ++ + ++ + +

++++

∆∆∆∆= + + −= + + −= + + −= + + − ∆∆∆∆

∆∆∆∆====

∆∆∆∆

Numerical Solution: Numerical Solution: second-order forward upstream (Tremback et al 1987)

adv

s

t

∂

∂

SubSub--grid Diffusion Transportgrid Diffusion Transport

Diffusion term:Diffusion term:

(((( ))))0

0

1 i

PBL i iturb

u ss

t x

ρρρρ

ρρρρ

′ ′′ ′′ ′′ ′∂∂∂∂∂∂∂∂ = −= −= −= − ∂ ∂∂ ∂∂ ∂∂ ∂

∑∑∑∑

Numerical Solution: Numerical Solution: unresolved transport using K-theory in which the covariances are evaluated as the product of an eddy mixing coefficient and the gradient of the transported mean quantity:

ii h

i

su s K

x

∂∂∂∂′ ′′ ′′ ′′ ′ = −= −= −= −

∂∂∂∂

Diffusion coefficients need to be specified as a function of flow characteristics (e.g. shear, stability, length scales). boundary layer eddies

zinv

Biomass burning emissions inventoryBiomass burning emissions inventoryBrazilian Fire Emission Model: Brazilian Fire Emission Model: Regional scale Regional scale –– daily basisdaily basis

[ ][ ] . . . ,veg

f fireveg vegM aηη α β= Ε

density of carbon data

land use data

near real time fire product

emission & combustion factors

mass estimation

CO source emission (kg m-2day-1)Freitas et al, 1999, 2005

SubSub--grid convective transport term:grid convective transport term:(coupled to the cumulus parameterization)(coupled to the cumulus parameterization)

(((( ))))0

0

1

conv

w ss

t z

ρρρρ

ρρρρ

′ ′′ ′′ ′′ ′∂∂∂∂∂∂∂∂ = −= −= −= −

∂ ∂∂ ∂∂ ∂∂ ∂

[[[[ ]]]] [[[[ ]]]], ,

updraft / downdraft flows

mass flux where the flows originate / normalized mass flux profile

: in cloud value of scalar/

enviromen

( , ) ( , ) ( , ) ( , )

, :

, :

:

u u u b u d d d b d

u d

w s s s z m z d s s z m z d

u d

m

s

s

λ λλ λλ λλ λ

η λ λ λ η λ λ λη λ λ λ η λ λ λη λ λ λ η λ λ λη λ λ λ η λ λ λ

ηηηη

′ ′′ ′′ ′′ ′ = − − −= − − −= − − −= − − −∫ ∫∫ ∫∫ ∫∫ ∫% %% %% %% %

%%%%

represents integral over all present clouds in the model grid box

t value of scalar

model value

:

:s

λλλλ∫∫∫∫

Mass flux approachMass flux approach

SubSub--grid grid deepdeep convective transport term:convective transport term:

[[[[ ]]]] [[[[ ]]]] ( )( ) ( )

( )

( )

( )

( )

d

deepco

b

bdeepo

v

c nv

n

u u u d

u u

d d

u b

d d

u d

d d

u

z

z

z z

z

s s m s s m

m z

m z

m m

w ss s s

M

m

w s η ηη ηη ηη η

ε ηε ηε ηε η

εεεε

η ηη ηη ηη η′ ′′ ′′ ′′ ′

= − −= − −= − −= − −

= − − −= − − −= − − −= − − −

====

= += += += +

′ ′′ ′′ ′′ ′

% %% %% %% %

%%%%

%%%% %%%%

defining

using

2D spectral model: deep and shallow (non2D spectral model: deep and shallow (non--precipitating) cumulusprecipitating) cumulus

case of deep cumuluscase of deep cumulus

Parameterized deep convective transportParameterized deep convective transport

updraft detrainmentupdraft detrainment downdraft detrainmentdowndraft detrainmentenvironment environment subsidencesubsidence

( )

( ) ( )0

and using the mass conservation equation

( )

( )

1

%%

%

%%

%deepconv

b

b

u

u u udeepco

u d d

u

nv

ud d d

z

z

s s

z

w ss s

s st

z z

s s

sm

mδ η

ρ

η εη η

δ η η∂

= − + + ∂

∂

∂

′ ′∂ ∂=

−

∂

−∂

−

The closure problemThe static control and entrainment /detrainment assumptions

determine the vertical structure of the tracer transport, however the determination of the overall magnitude of the transport requires the determination of the mass flux at cloud base : mu(zb)

We use new Grell´s cumulus scheme that provides this number using an ensemble version of closures (Kuo, Arakawa&Schubert, Grell, Kain&Fritsch, Brown).

Deep convective transport of CODeep convective transport of CO21Z 24 Sep 200221Z 24 Sep 2002

Vertical section at lat 10SVertical section at lat 10S

vertical vertical

levellevel

11.5 km 11.5 km

CO (ppb)CO (ppb)

Shallow convective transportShallow convective transportof smoke/gases from biomass burningof smoke/gases from biomass burning

SMOKESMOKE

Flows in shallow convective cloudsFlows in shallow convective clouds

environment

subsidence

~ 5 km

updrafts

cloud base

cloud top

boundary layer inflow

Parameterized shallow convective transportParameterized shallow convective transport

•• Based at the mass flux cumulus schemeBased at the mass flux cumulus scheme (Grell 1993; Grell and Devenyi 2002).

•• Transport termTransport term (donwdrafts are disregarding)

We use a stability closure (like Kain & Fristch) to determine the mass flux at shallow cloud base

(((( ))))0

b

u u ushallowconv

s ss s

t z

mδ ηδ ηδ ηδ η

ρρρρηηηη

∂∂∂∂∂∂∂∂ ====

∂∂∂∂− +− +− +− + ∂∂∂∂

%%%%%%%% %%%%

updraft detrainmentupdraft detrainment environment environment subsidencesubsidence

Shallow convective transportShallow convective transportModel simulation for 14-17Z03092003

SMOKESMOKE

shallow shallow cumuluscumulus

PBL

PBL

Low Low troposphere troposphere COCO

CO at 3300mCO at 3300m

Another important subAnother important sub--grid process,grid process,but frequently ignoredbut frequently ignored

PlumePlume--rise due to the strong buoyancy of the hot gases/aerosols emitterise due to the strong buoyancy of the hot gases/aerosols emittedd

8 km

RondôniaRondônia, 2002, 2002

Convective cloud Convective cloud

above the above the BorBor

Forest Island fire, Forest Island fire,

6 July 19936 July 1993..

How to include this subHow to include this sub--grid transport in the model?grid transport in the model?

1D cloud resolving model (CRM) using the 1D cloud resolving model (CRM) using the supersuper--pameterizationpameterization conceptconcept:

• Use a 1D CRM embedded in each column of the large-scale atmospheric-chemistry transport model;

• Each grid box with fires, pass the large-scale condition of the host model to the 1D CRM;

• Resolve explicitly the motion of the plume;• Return to the host model with the final rise of

the plume (or the injection layer);• Take account this plume rise at the source

emission, releasing material emitted at flaming phase at this layer.

massinflow

massoutflow

1D embedded model

Column of the 3d host

model: Tenv, rvenv, penv

Cartoon describing Cartoon describing

the methodologythe methodology

The 1D cloud resolving model: The 1D cloud resolving model: governing equationsgoverning equations

2

1

1 0 5

1

1 2

2

2

2

e

microphysicsp

v v vv v

c

e

microphysics

Simpson&Wiggert, 1968

Siebesma et al, subm. J

w ww gB w

t z R

T T g Tw w w T T

t z c R t

r r rw w r r

t

t

t

r

z R

S

w

A

.

( )

( )

γγγγ

γγγγµµµµ

ααααγγγγ

αααα

αααα

∂ ∂∂ ∂∂ ∂∂ ∂+ = −+ = −+ = −+ = −

∂ ∂∂ ∂∂ ∂∂ ∂

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂ + = − − − ++ = − − − ++ = − − − ++ = − − − +

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂ + = −+ = −+ = −+ = −

==== ++++

− +− +− +− +

====−−−−

∂ ∂∂ ∂∂ ∂∂ ∂++++

∂∂∂∂

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂

2

2

c c

c

microphysics

ice rain ice rain ice rain

ice rain

micro

v c rain ice

microphysics

physics

Kessler, 1969

Ogur

bulk microphysics:

T r r r r t

r rw r

z R t

r r rw w r

t z R t

, , ,

,

( , , , ,

sedim

), sedim

αααα

ξξξξ

αααα

ξξξξ

∂∂∂∂

∂∂∂∂ ==== ∂∂∂∂

= − += − += − += − + ∂ ∂∂ ∂∂ ∂∂ ∂

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂ + = − + ++ = − + ++ = − + ++ = − + +

∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂∂ ∂ ∂

a & Takahashi,1971

Berry,1967

dynamicsdynamics

thermodynamicsthermodynamics

water vapor water vapor conservationconservation

bulk microphysicsbulk microphysics

cloud water cloud water conservationconservation

rain/ice rain/ice conservationconservation

The lower boundary conditionThe lower boundary condition

the closurethe closure

Morton, Taylor & Turner (1956):

"Turbulent grav. convection from maintained and instantaneous sourc

buoyancy flux

plume

es

radi

u

"

s 6

5

(

ππππ

αααα

ℜ Εℜ Εℜ Εℜ Ε====

====

p e

gF A

c p

R z

w z

(((( ))))

boundary condition for

density correction

boundary condition for

where: entrainment coeffic

ient,

1/3

5/3

1/3

5 0.9

6

5

6 0.9

0.2

0.

) ( )

( )( )

1

αααα

αααα

αααα αααα

αααα

ρρρρ

ρρρρ

ρρρρ

ρρρρ

−−−−

====

∆∆∆∆====

====∆∆∆∆

−−−−

====

====

v

v

v

e

e v

e

v

v

F

F

z

zF

g

T z

w

z

T z T

virtual boundary height19αααα−−−−

surfR

-17

-2convective energy from fire (Wm )

(McCarter & Broido, 1965)

1.5 to 2.1 10

(heat flux) fuel load / combustion factor

plume area instantaneous fire size

joules kg

flux

flux

A

E

0.4 0.8

h

tc

E

h cββββββββ

≡ ≈≡ ≈≡ ≈≡ ≈

≡≡≡≡

− Ξ− Ξ− Ξ− Ξ

ΞΞΞΞ

====

≅≅≅≅

==== ====∆∆∆∆

∆∆∆∆ flaming phase duration

(water flux

flux ) .5 c

t

W 0 ββββ

====

====

PlumePlume--rise of vegetation fires:rise of vegetation fires:typical energy fluxes (kWmtypical energy fluxes (kWm--22))

3.3

23.

80.

Upper bound kWmkWm--22

97%Pasture - grassland cropland

75%4.4Woody savanna -cerrado

45%30.Tropical forest

Flaming consumption

Lower bound kWm-2

Biome type

Refs: Carvalho et al, 1995-2001-2005 (com. pessoal);

Riggan et al, 2004;

Ward et al, 2002;

Ferguson et al, 1998;

Cochrane et al; 200X-com. pessoal;

Miranda et al, 1993.

Including plume rise mechanism troughIncluding plume rise mechanism troughsupersuper--parameterizationparameterization conceptconcept

280tot

parcel env

al

Th kWm

rw

T−−−−

−−−− ====

230 total

parcel envh kWm

T T

rw

−−−−

====

−−−−

2

230

80

w h kWm

w h kWm −−−−

−−−−⇐ =⇐ =⇐ =⇐ =

⇐ =⇐ =⇐ =⇐ =

Injection Injection layerlayer

plume top

lower bound

plume topupper bound

Freitas et al., 2006 GRL under review

1D plume1D plume--rise model for vegetation fires rise model for vegetation fires Biome: Biome: ForestForest

Time duration: 50 mnFire size: 20 ha

Heat flux: 80 kWm-2 / 30 kWm-2

Example of CO source emission field with the plumeExample of CO source emission field with the plume--rise for rise for vegetation fires at the CATTvegetation fires at the CATT--BRAMS host modelBRAMS host model

South AmericaSouth Americamostly forest firesmostly forest fires

AfricaAfricamostly savanna firesmostly savanna fires

flamingflamingemissionemission

smolderingemission

( )

0

1

,, , ,

2

,

,

*

0,

0

onde

é a velocidade de sedimentação

é a resistênci

2 ( )

9

a aerodinâmica

ln

Resistance Formulation

r

i p a

f i i

a

z

hz q

a

d part i a b a b f i i

b

f

m

d

v R

r gV G

dz

z

R R R V V

z

v

Ru

R

F S

z

ρ ρ

η

φ

κ

−

−=

= −

+

=

=

= + +

∫

( )2

3

, *

é a resistência da Pr

difusão molecul arq

Sc

ku

Dry deposition and particles sedimentation Dry deposition and particles sedimentation fully coupled with LEAFfully coupled with LEAF--3, 3,

including the patches approachincluding the patches approach

BiomassBiomass burningburning AerosolAerosol modelmodel(Procópio (Procópio etet al., 2003)al., 2003)

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

200 600 1000 1400 1800 2200 2600 3000

wavelength (nm)

para

mete

rs ωω ωω

o a

nd

g

0

0.5

1

1.5

2

2.5

3

3.5

4

para

mete

r Q

e

Qe; ; ; ; ττττ<0.20

Qe; 0.20<ττττ<1.60

Qe; ττττ>1.60

g; ; ; ; ττττ<0.20

g; 0.20<ττττ<1.60

g; ττττ>1.60

Biomass burning aerosol model

ωωωω ο; ο; ο; ο; ττττ<0.20

ωωωωo; 0.20<ττττ<1.60

ωωωωo; ττττ>1.60

Lookup table with Qe, ωωωω0000 and g

as function of ττττ(500 nm) and λ

1,6

1,4

1,2

1,0

0,8

0,6

0,4

0,2

0

τ(500nm)

Interação da radiação com a Interação da radiação com a microfísica e convecçãomicrofísica e convecção

((SunSun & & ShineShine, , SavijärviSavijärvi etet al.) al.) –– ECMWF, ECMWF, MesoMeso--NHNH

c r L I r

0 0L L 0I I 0r r

0L L L 0I I I 0r r r 0

( )

( ) /

g ( g g g ) /

τ = τ + τ = τ + τ + τ

ω = ω τ + ω τ + ω τ τ

= ω τ + ω τ + ω τ ω

Propriedades ópticas de nuvens

e da chuva

AOT 550 AOT 550 nmnm((fromfrom biomassbiomass burningburning))

Solar Solar RadiationRadiationatat surfacesurface ((WmWm--22))

Algumas validações do modeloAlgumas validações do modelo

Model validation using 2002 dataModel validation using 2002 data

Grid 2: 35 km

Grid 1: 140 km

• Model configuration: 2 grids with 2-way nesting technique.• Simulation time: 135 days starting 00Z15Jul2002.• Initial and boundary condition using 4DDA from CPTEC analysis fields.• Ensemble version of convective parameterization (Grell and Dévényi

(2002), deep and shallow) • Full microphysics, PBL diffusion with TKE closure, etc…• Soil moisture initialization (Gevaerd & Freitas, 2006).• Biomass burning emission using a hybrid fire remote sensing database

through combination of MODIS/AVHRR/GOES fire products.• EDGAR 2000FT for anthropogenic emissions.

ModelModel ValidationValidationwithwith AIRS AIRS 500 hPa CO500 hPa CO

ComparisonComparison betweenbetween AOT (550 AOT (550 nmnm))MODIS x MODELMODIS x MODEL

Comparison between CO columnComparison between CO column(10(1018 18 molecmolec cmcm--22) September, 2002 mean) September, 2002 mean

MOPITT MOPITT

(V3)(V3)

MODELMODEL

Model comparison with near SMOCC/RACCI surface Model comparison with near SMOCC/RACCI surface observations: observations: CO and PM2.5 observationsCO and PM2.5 observations in in RondôniaRondônia

Model comparison with CO airborne observations Model comparison with CO airborne observations from SMOCC campaign in from SMOCC campaign in RondôniaRondônia

Funcionalidades do CATT que migrarão Funcionalidades do CATT que migrarão para o BRAMS 4.0para o BRAMS 4.0

• Nova versão da parametrização convectiva de Grell (training, acoplamento do CAPMAX com propriedades da PBL, versão bi-espectral)

• Módulo de transporte para espécies gasosas insolúveis e aerossol

• Deposição/sedimentação de aerossóis e parametrização de vida-média para gases

• Radiação CARMA acoplado à microfísica/cumulus e com modelo de aerossol de queimadas

• Leaf -3 com perfil de raízes do Arora&Boer E.I. 200X e com NDVI baseado no MODIS 2000/2001

• Nova umidade do solo utilizando dados mais apropriados para solos tropicais (H&T) seguindo o formalismo de van Genutchen

NDVI eNDVI eLAILAI

http://www.cptec.inpe.br/brams/input_datahttp://www.cptec.inpe.br/brams/input_data..shtmlshtml