3-D global simulations of tropospheric CO distributions...
Transcript of 3-D global simulations of tropospheric CO distributions...
3-D global simulations of tropospheric CO distributions ± resultsof the GIM/IGAC intercomparison 1997 exercise
M. Kanakidou a,*, F.J. Dentener b, G.P. Brasseur c, T.K. Berntsen d, W.J. Collins e,D.A. Hauglustaine c,f, S. Houweling b, I.S.A. Isaksen d, M. Krol b, M.G. Lawrence g,
J.-F. Muller h, N. Poisson a,1, G.J. Roelofs b, Y. Wang i,2, W.M.F. Wauben j
a Laboratoire des Sciences du Climat et de lÕEnvironnement, Unite Mixte CNRS/CEA, Orme des Merisiers, CE Saclay,
F-91191 Gif-sur-Yvette Cedex, Franceb Utrecht University, IMAU, Utrecht, The Netherlands
c Atmospheric Chemistry Division ± NCAR, Boulder, CO, USAd University of Oslo, Oslo, Norway
e Atmospheric Chemistry Modelling, Meteorological O�ce, Bracknell, UKf Service dÕAeronomie, Universit�e Paris 6, Jussieu, Paris, France
g Max Planck Institute for Chemistry, Atmospheric Chemistry Division, Mainz, Germanyh Belgian Institute for Space Aeronomy, Brussels, Belgium
i Harvard University, Cambridge, MA, USAj KNMI, De Bilt, The Netherlands
Received 17 April 1998; accepted 7 January 1999
Importance of this Paper: The current generation of 3-D global models used for chemistry and climate global simulations
have been compared in the frame of the Tropospheric Ozone (O3) Global Model Intercomparison Exercise performed in
1997. The objective was to systematically evaluate their capabilities to simulate tropospheric ozone and their precursor
gases, including carbon monoxide, and to identify key areas of uncertainty in our understanding of the tropospheric O3
budget. This exercise was organised by Global Integration Modelling (GIM) Activity of the International Global At-
mospheric Chemistry (IGAC) Project.
Abstract
The objective of the Tropospheric Ozone (O3) Global Model Intercomparison Exercise performed in 1997 was to
systematically evaluate the capabilities of the current generation of 3-dimensional global models to simulate tropo-
spheric ozone and their precursor gases, and to identify key areas of uncertainty in our understanding of the tropo-
spheric O3 budget. This exercise has been organised by Global Integration Modelling (GIM) Activity of the
International Global Atmospheric Chemistry (IGAC) Project. The present paper focuses on the capability of the
Chemosphere: Global Change Science 1 (1999) 263±282
* Corresponding author. Department of Chemistry, Environmental Chemical Process Laboratory, University of Crete, P.O. Box
1470, 71409 Heraklion, Greece. Tel.: +30-81-393-633; fax: +30-81-393-601; e-mail: [email protected] Present address: Harvard University, Cambridge, MA, USA.2 Present address: Georgia Institute of Technology, Atlanta, USA.
1465-9972/99/$ ± see front matter Ó 1999 Elsevier Science Ltd. All rights reserved.
PII: S 1 4 6 5 - 9 9 7 2 ( 9 9 ) 0 0 0 2 9 - X
models to simulate carbon monoxide, which is an important pollutant in the troposphere. The intercomparison of
twelve 3-dimensional global chemistry/transport models shows signi®cant di�erences between the models although all
of them capture the general patterns in the global distribution of CO. Ó 1999 Elsevier Science Ltd. All rights reserved.
Keywords: 3-dimensional; Global; Models; Intercomparison; Tropospheric chemistry; IGAC; Carbon monoxide; CO;
Budget; Methane lifetime
1. Introduction
It is now recognised that the atmospheric con-centrations of several chemical and radiativelyimportant trace constituents (gases and particles)are changing in the atmosphere (IPCC, 1994).Human activity is the main reason for thesechanges, climate changes a�ecting natural emis-sions and the radiative balance of the earth beingthe second forcing agent. Indeed, several traceconstituents like aerosols, methane (CH4), nitrousoxide (N2O), nitrogen oxides (NO + NO2), carbonmonoxide (CO), Non-Methane Volatile Com-pounds (NMVOC) including Non-Methane Hy-drocarbons (NMHC), sulphur dioxide (SO2),dimethylsulphide (DMS) and halocarbons havedirect anthropogenic and/or natural emissions tothe atmosphere. Others, like ozone (O3) and sec-ondary aerosols, are produced from chemical re-actions initiated by their precursors. In turn, thechanging atmospheric concentrations of thesetrace constituents a�ect the radiative balance ofthe earth and may lead to a climate change.
To understand and reliably predict chemicaland climate changes in the atmosphere a thoroughunderstanding of the chemical, physical, biologicaland climatic processes which a�ect the distribu-tions of trace compounds in the atmosphere, andof the interactions between these processes is re-quired. Only then, the global budgets of traceconstituents, their evolution due to natural andanthropogenic forcing and the related feedbackmechanisms can be realistically simulated by themean of numerical models.
The use of numerical models is required to takeinto account the physical processes governing thebudget of trace gases in the troposphere and thecomplex non-linear behaviour of chemical reac-tions in the troposphere. Such models integrate themost recent information on chemical kinetics of
homogeneous and heterogeneous reactions as wellas photodissociation coe�cients of atmosphericmolecules. They also account for interactions be-tween transport, chemistry and the radiative bud-get in the troposphere, and speci®cally for: (i)transport of trace constituents like nitrogen species(NOy) and O3 from the stratosphere, (ii) emissionsof trace gases like O3 precursors: nitrogen oxides,carbon monoxide, NMVOCs and aerosols occur-ring mainly in the low troposphere, (iii) dry de-position of various trace gases and particles on theearthÕs surface, (iv) wet removal of soluble gasesand particles, (v) convection, (vi) planetary boun-dary layer mixing, (vii) synoptic mixing in thetroposphere, (viii) cloud microphysics and (ix)homogeneous and heterogeneous chemical reac-tions.
Recent assessments of the global-scale, radia-tive e�ects of greenhouse gases and of aerosolshave been performed on the basis of global modelswhich, in most cases, are still under development.The accuracy of such calculations relies stronglyon the capability of the models to simulate theglobal budget of trace gases and aerosols. A majorconcern for the calculation of the chemically re-active trace gases budget is associated with: (1) theinaccuracies in the transport parameterisation(Jacob et al., 1997), (2) the simpli®cation of thechemistry in the troposphere, required due to theprohibitive number of chemical reactions andspecies and (3) the uncertainties in the trace gasesemissions used in the global models.
Even though current models ignore some im-portant chemical and radiative processes in theatmosphere, it seems crucial for the time being toimprove our model estimate of the budgets of O3,CH4, CO and NOx. Indeed, O3 is a greenhouse gas,a pollutant and one of the major oxidising agentsin the troposphere. CH4 is the most chemicallyimportant greenhouse gas with direct emissions to
264 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
the atmosphere. NOx, which is also a pollutant,and in particular NO and NO2, controls O3
chemical production and destruction in the tro-posphere. CO is a chemically important gas withprimary and secondary emissions in the atmo-sphere. Depending on the ambient NOx mixingratios, CO oxidation by hydroxyl (OH) radicalsforming hydrogen peroxy radicals (HO2) enhances(high NOx) or depletes (low NOx) ozone in thetroposphere (Crutzen, 1994). CO changes havefeedbacks into the abundance of CH4 through thealteration of OH (Prather, 1996). Thus, althoughCO itself is not a signi®cant greenhouse gas,changes in its concentration a�ect ozone and CH4,both signi®cant greenhouse gases (Daniel andSolomon, 1998).
E�ort has already started by WMO/IPCC in1994 (Stordal et al., 1995; Olson et al., 1997) toevaluate the capability of the chemical models tosimulate O3, NOx, HOx and VOC chemistry. Thise�ort is continued by the Global IntegrationModelling (GIM) Activity within the InternationalGlobal Atmospheric Chemistry (IGAC) Project ofIGBP in 1996±1997, and focuses on the capabilityof the global 3-D chemistry/transport models usedfor global chemical and climate simulations tocalculate the O3 budget. Ground-based and in situobservations from several regions of the globe areused for validation of the model results. Thepresent paper focuses on the model capability tosimulate carbon monoxide since it is a key com-pound for tropospheric chemistry. Moreover,carbon monoxide has a more simple and betterunderstood chemistry and better documented di-rect sources than the shorter lived NOx and VOCs.It is therefore a good tool to evaluate the chemistryand the transport characteristics of photochemicalmodels. Ozone simulations are the topic of anotherpaper in preparation.
2. The GIM/IGAC tropospheric ozone chemistry/
transport 3-D models intercomparison exercise
The objective of the Tropospheric OzoneGlobal Model Intercomparison Exercise perform-ed in 1997 was to systematically evaluate the ca-pabilities of the current generation of 3-D global
models to simulate tropospheric O3 and otherchemical compounds, and to identify key areas ofuncertainty in our understanding of the tropo-spheric O3 budget. This exercise has been organ-ised by GIM Activity within IGAC Project. Thestrategy was to investigate the consistency of thepredicted results between models, which use a vari-ety of inputs and mechanisms, to compare modelresults to the real atmosphere and identify areas forfurther research in the close future.
To conduct this exercise, several modellinggroups were asked to provide the Ôbest caseÕ outputof their models, and speci®cally:· O3, CO and NOx (NO + NO2) monthly mean
mixing ratios at surface, 500 and 200 mbar inJanuary and July;
· surface monthly mean concentrations at Bar-row, Mace Head, Hohenpeissenberg, MaunaLoa, Barbados Samoa, Cape Grim, Cape Pointfor O3; and at Barrow, Mace Head, MaunaLoa, Samoa, Cape Grim for CO;
· monthly varying vertical O3 pro®les at selectedozone sonde stations (Resolute, Hohenpeissen-berg, Wallops island, Bermudas, Hilo, Natal,Samoa, Lauder, Syowa, S. Pole);
· seasonal and annual mean terms of O3 budgetup to 300 mbar for the 4 seasons annual meanCH4 lifetime and mean hemispheric and globalburdens.
3. Description of the chemistry transport models
The 12 global 3-D Chemistry Transport Modelsthat participated at this exercise are listed in Table1 together with the contributing scientists and keyreferences. All these models are global 3-D models.Three of them are climatological models, i.e., theysimulate atmospheric transport based on pre-cal-culated or observed monthly mean winds andtemperatures. One of the models is a general cir-culation model (ECHAM) and the remainingmodels are chemistry/transport models using pre-calculated/observed meteorology provided every4±12 h. This meteorological input is derived fromgeneral circulation models or weather predictionmodels. Models also di�er in the parameterisationof the main processes controlling chemical tracer
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 265
budgets, i.e. transport by advection, di�usion andconvection, chemistry (homogeneous and hetero-geneous), wet and dry deposition and emission oftrace compounds (including CO) by natural andanthropogenic sources. Detailed model descrip-tions can be found in the literature (see referencesin Table 1).
Some of the model characteristics together withthe annual global emissions of ozone precursors:NOx, CO, CH4 and NMVOC used for the Ôbestcase present dayÕ simulations are summarised inTables 2 and 3, respectively. Note that the directannual amount CO emissions adopted in eachmodel can be signi®cantly di�erent: Total primarysources vary from 1040 to 2362 Tg-CO/yr amongthe various models, as shown in Table 3. Theseemissions are primarily associated with techno-logical sources, in the northern hemisphere, andwith biomass burning, in the tropics. In somemodels, those not explicitly considering NMVOCchemistry, CO pseudo-primary emissions are usedto represent the photochemical source fromNMVOC oxidation. However, the major reasonfor the large range of CO emissions is linked to theuncertainties in the global CO emission estimatesdeduced from smaller scale observations. The at-mospheric dynamics (Table 2a) and the spatial andtemporal resolutions of each model are di�erent,implying di�erences in the propagation of theemissions in each model. Finally, the chemicalschemes adopted for NMVOC chemistry repre-
sentation (Table 2b) and consequently the leveland distributions of the hydroxyl (OH) radical andthe photochemical production of CO also di�er ineach model. In addition, one should note that thenumerical solvers used to integrate the chemicaldi�erential equations are di�erent between models.
One model (GFDL) which participated in theoverall intercomparison exercise, uses paramete-rised chemistry and imposes pre-calculated COdistributions. Therefore the results and the char-acteristics of this particular model will not beconsidered in the present paper. All other modelsprovided:· the calculated CO monthly mean mixing ratios
from the best case simulation at the EarthÕs sur-face, 500 and 200 mbar in January and July.
· the CO monthly mean concentrations (andstandard deviation) at 5 selected surface sta-tions (or the grid points of the models closestto these stations) where observations are avail-able.
4. Results and discussion
The 3-D CO distributions calculated by the 11models show important di�erences as demon-strated in Figs. 1±4 for January and Figs. 5±8 forJuly ((a) at surface and (b) at 500 hPa). CalculatedCO distributions re¯ect CO direct emissions,photochemical production by NMVOC and CH4
Table 1
3-D models that participated at the Tropospheric Ozone (O3) Global Integration Modelling (GIM) intercomparison exercise (3-D
models)
IMAGES J.-F. Muller, G. Brasseur, C. Granier Muller and Brasseur, 1995
GFDL H. Levy Kasibhatla et al., 1996
HARVARD Y. Wang, D. Jacob Wang et al., 1998a,b,c
ECHAM G.-J. Roelofs Roelofs and Lelieveld, 1995, 1997
TM3 F. Dentener, S. Houweling Houweling et al., 1998
IMAU3 M. Krol Krol and Weele, 1997
CTMK W.M.F. Wauben Wauben et al., 1997, 1998
MATCH M.G. Lawrence, P.J. Crutzen Lawrence, 1996, Lawrence et al., 1995, Lawrence
and Crutzen, 1998
MOGUNTIA N. Poisson, M. Kanakidou Poisson, 1997, Poisson et al, 1999
MOZART D.A. Hauglustaine, G.P. Brasseur, S. Walter Brasseur et al., 1996, 1998
UKMETO R.G. Derwent, C.E. Johnson, W.J. Collins, D.S.
Stevenson
Collins et al., 1997
UIO T.K. Berntsen, I.S.A. Isaksen Bernsten and Isaksen, 1997
266 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
Tab
le2
MO
DE
LIM
AU
3IM
AG
ES
HA
R-
VA
RD
UK
-
ME
TO
EC
HA
MC
TM
KM
AT
CH
MO
-
ZA
RT
TM
3M
OG
UN
TIA
UIO
Last
sub
mis
sio
n
of
da
tao
n
Ap
ril
97
Ap
ril
97
Oct
ob
er
97
Oct
ob
er
97
Marc
h97
Ap
ril
97
Ap
ril
97
Sep
tem
-
ber
97
Marc
h97
Oct
ob
er97
Ap
ril
97
(a)
Mo
del
des
crip
tio
n
GC
M/C
TM
(Y/N
)N
ON
ON
ON
OY
ES
NO
NO
NO
NO
NO
NO
Syn
op
tic
NO
NO
YE
SY
ES
YE
SY
ES
YE
SY
ES
YE
SN
OY
ES
Up
per
bo
un
da
ry1
00
mb
50
mb
10
mb
100
mb
10
mb
5m
b2
mb
5m
b10
mb
100
mb
10
mb
Nu
mb
ero
f
ver
tica
lla
yer
s
10
25
9(1
9)
919
15
28
25
19
10
9
Iso
ba
ric/
sigm
aIs
ob
ari
cS
igm
aS
igm
aH
yb
rid
Sig
ma
Sig
ma
Sig
ma
Hyb
rid
Hyb
rid
Iso
bari
cS
igm
a
Ho
rizo
nta
l
reso
luti
on
10
°´1
0°
5°´
5°
4°´
5°
5°´
5°
3.7
5°
´3.7
5°
8°´
10°
1.9
°´1.9
°2.8
°´2.8
°3.7
5°
´3.7
5°
10°´
10
°8°´
10°
(b)
Ch
emis
try
des
crip
tio
nin
the
mo
del
s
NM
VO
Cch
emis
try
NO
aY
ES
YE
SY
ES
NO
aN
Oa
NO
aY
ES
YE
SY
ES
YE
S
±N
oo
fN
MV
OC
b0
66
11
00
07
77
6
Ch
emic
al
solv
erQ
SS
AQ
SS
Ap
ara
met
e-
rise
d
EB
IE
BI
QS
SA
QS
SA
EB
IE
BI
QS
SA
QS
SA
Tim
este
pfo
r
chem
istr
y
2h
6h
c4
hd
5m
in30
min
2h
30
min
20
min
40
min
2h
30
min
Ph
oto
lysi
sra
tes
Ca
lcu
late
dT
ab
les
Pa
ram
ete-
rise
d
ba
sed
on
6st
ream
2st
ream
2st
ream
Tab
les
2st
ream
Tab
les
Tab
les
Tab
les
2st
ream
Fre
qu
ency
of
rea
d-
ing
or
calc
ula
tin
g
ph
oto
lysi
sra
tes
c4
h45
min
5d
15
days
Ever
y30
min
20
min
15
d15
d1
mo
aC
Ose
con
da
ryso
urc
eis
ad
ded
as
ap
seu
do
-dir
ect
sou
rce;
EB
I:E
ule
rian
Back
ward
,Q
SS
A:
Qu
asi
Ste
ad
y-S
tate
Ass
um
pti
on
;D
t:ti
mes
tep
.b
Incl
ud
ing
carb
on
yls
for
wh
ich
dir
ect
emis
sio
ns
are
con
sid
ered
inth
em
od
el.
cA
tth
eb
egin
nin
go
fea
chm
on
tho
fsi
mu
lati
on
tim
este
pis
30
min
for
the
®rs
tth
ree
days.
dA
fou
rth
ord
erin
teg
rati
on
sch
eme
isu
sed
.
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 267
Tab
le3
Em
issi
on
so
fo
zon
ep
recu
rso
rsu
sed
inth
eC
TM
s
MO
DE
LIM
-
AU
3
IMA
G-
ES
(I)
HA
R-
VA
RD
UK
-
ME
TO
EC
-
HA
M
CT
MK
MA
T-
CH
MO
-
ZA
RT
TM
3M
O-
GU
N-
TIA
UIO
Ran
ge
CO
dir
ect
emis
sio
ns
(Tg
-CO
/yr)
13
50
12
83
1040
1168
1250
1427
1680
a1218
1110
1185
2362
b1040±2362
Fu
elco
nsu
mp
tio
n,
ind
ust
ry5
40
47
1520
303
450
383
560
a382
400
500
893
Bio
ma
ssb
urn
ing
49
06
26
520
725
700
716
920
a662
450
515
714
Veg
eta
tio
n/s
oil
28
01
66
NO
100
100
166
150
a162
75
c130
387
Oce
an
s4
02
0N
O40
162
50
13
40
40
378
CO
seco
nd
ary
sou
rce
10
63
14
59
1090
1320
840
dN
C884
d;a
NC
1350
1235
NC
840±1459
CO
surf
ace
loss
NC
26
4N
O270
NO
NC
233
NC
320
330
NC
0±330
NO
emis
sio
ns
(Tg
-N/y
r)3
7.1
36
.85
45
47.5
37.5
47.6
939
39.9
538.4
38
42.7
236.8
±47.7
Fu
elco
nsu
mp
tio
n,
ind
ust
ry2
1(G
)2
121
(G)
21
(G)
21
22
21.3
(G)
21.4
(G)
22
(G)
21.3
(G)
21
(G)
Bio
ma
ssb
urn
ing
6e
4.9
12
96
e5.3
e8.5
e4.5
5e
6e
4.4
So
il4
.85
.56
12
f5.5
3.9
5.5
f6.6
5.5
f5.5
12.2
g
Lig
htn
ing
4.7
53
5h
55
2.2
h7
5.3
55.5
Air
cra
ft0
.60
.45
0.5
0.5
NO
0.8
50.5
0.4
50.6
NO
0.5
2
Tra
nsp
ort
fro
mst
rato
sph
ere
NC
NC
0.5
i0.4
5N
C0.6
41.0
NC
NC
0.2
NC
CH
4em
issi
on
s(T
g-C
H4/y
r)5
23
48
5460
j452
kj
540
j469
j540
304
304±540
NM
VO
Cem
issi
on
s(T
g-C
/yr)
NO
l1
25
4680
442.4
650
NO
NO
741
520
748.8
442±1254
An
thro
po
gen
icN
O1
75
80
58.4
330
(CO
)
NO
225
a
(CO
)
125
(EG
)120
114
m
Na
tura
lN
O1
07
9600
384
320
(CO
)
NO
350
a
(CO
)
516
400
400
+-
150
634.8
m
a8
80
Tg
CO
/yr
an
thro
po
gen
icb
urn
ing
sa
nd
40
Tg
CO
/yr
wil
d®
res;
225
Tg
CO
/yfr
om
oxid
ati
on
of
NM
VO
Cfr
om
tech
no
logic
al
sou
rces
;350
Tg-C
O/y
fro
mn
atu
ral
NM
VO
Co
xid
ati
on
.b
Glo
ba
lv
alu
efo
rmH
ou
gh
(19
91)
an
dg
eog
rap
hic
al
dis
trib
uti
on
as
for
NO
x,C
Od
ryd
epo
siti
on
vel
oci
ty�
0.0
3cm
/so
ver
lan
d.
cA
llb
iosp
her
e.d
Fro
mC
H4
ox
idati
on
on
ly,
for
EC
HA
Mb
elo
w3
00
mb
ar.
eD
istr
ibu
tio
nta
ken
fro
mH
ao
eta
l.(1
99
0).
fY
ien
ger
an
dL
evy
(19
95).
gB
iom
ass
bu
rnin
g+
soil
s.h
Pri
cea
nd
Rin
d(1
99
2);
Law
ren
ceet
al.
(19
95).
iS
um
of
NO
xa
nd
HN
O3.
jP
resc
rib
edsu
rface
con
cen
tra
tio
ns
for
CH
4a
cco
rdin
gto
Fu
ng
etal.
(1991).
kP
resc
rib
edsu
rface
con
cen
tra
tio
ns
of
17
72
pp
bv
NH
an
do
f1680
pp
bv
SH
.lIn
clu
des
on
lyC
2H
6a
nd
C6H
8ch
emis
try
,a
lso
dir
ect
CH
2O
emis
sio
ns
300
Tg/y
r.
(G)
GE
IAin
ven
tory
;(E
G)
ED
GA
R/G
EIA
inv
ento
ry;
(I)
Mu
ller
an
dB
rass
eur
(1995),
up
date
d;
Tu
rman
an
dE
dgan
;(N
C)
no
tca
lcu
late
d;
(NO
)n
ot
use
d;
(CO
)em
issi
on
s
intr
od
uce
din
the
form
of
carb
on
mo
no
xid
e;m
An
thro
po
gen
icem
issi
on
sa
cco
rdin
gto
EP
A;
na
tura
lem
issi
on
sacc
ord
ing
toIs
ak
sen
an
dH
ov
(1987).
268 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
Fig. 1. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of January at the surface
(1000 mbar).
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 269
Fig. 2. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of January at the surface
(continue).
270 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
Fig. 3. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of July at the surface (1000
mbar).
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 271
Fig. 4. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of July at the surface
(continue).
272 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
Fig. 5. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of January at 500 mbar.
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 273
Fig. 6. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of January at 500 mbar
(continue).
274 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
Fig. 7. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of July at 500 mbar.
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 275
Fig. 8. CO concentrations (ppbv) as computed by the eleven 3-dimensional global models for the month of July at 500 mbar (con-
tinue).
276 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
oxidation and destruction by the OH radical, aswell as advective and convective transport awayfrom the source regions and losses to the surface.
4.1. Surface CO distributions
Maximum surface CO concentrations are cal-culated by all models over technologically devel-oped areas of the world and over biomass burningregions. The calculated maximum CO concentra-tions depend on the model resolution. The low-resolution models are arti®cially more di�usivethan the higher resolution models by splittingemissions in their relatively large grid boxes. Fur-thermore, the vertical resolution of the surfacelayers is also di�erent between models and rangesbetween 50 and 400 m. The models with theshallowest surface layers generally show the high-est concentrations. Convective redistributions ofemissions, chemistry and surface loss strongly af-fect the ultimate surface concentrations.
Comparisons between Figs. 1 and 2 and Figs. 3and 4 (surface for January and July, respectively)show that lower CO concentrations are calculatedin July than in January in the northern hemisphere(NH) due to the strong photochemical sink of COduring NH summertime although the photo-chemical production of CO is also enhanced dur-ing that season. This is not the case in biomassburning areas, where CO concentrations remainhigh although their maximum values are shiftedfrom the NH tropics in January to the southernhemisphere (SH) tropics in July. Almost all modelscapture this pattern. However, signi®cant di�er-ences exist in the magnitude of CO seasonal cycleand absolute levels as shown in Figs. 1±4. Thesedi�erences will be further discussed when com-paring model results to observations.
Sharper interhemispheric gradients are calcu-lated for NH winter since CO increases in the NHdue to technological emissions (about 95% in thishemisphere) and relatively low photochemical de-struction under wintertime conditions. The longerlifetime of CO during winter than during summerallows CO to be transported over long distances,so that large concentrations of CO are calculatedduring winter downwind of the CO source areas.On the contrary, in July, high CO concentrations
are more restricted to source regions. These pat-terns are calculated by all models.
All models capture the general pattern of COlinked to the distribution of its surface sources andphotochemical production and destruction of CO.However, the calculated CO levels vary signi®-cantly between models due to (i) the annualamount of emissions adopted by each model (Ta-ble 3) as discussed above, (ii) the formulation ofmodel dynamics (Table 2a), (iii) the spatial reso-lution, and (iv) the chemical schemes (Table 2b)resulting in signi®cant di�erences in the hydroxyl(OH) radical distributions calculated by the mod-els. As a consequence the photochemical produc-tion of CO from VOCs and its photochemical sinkby reaction with OH radical di�er signi®cantlyfrom model to model (Table 3). Di�erences in theOH levels are re¯ected (i) on the photochemicalsource of CO that varies between 840 Tg-CO/yrand 1459 Tg-CO/yr and (ii) on the methane (CH4)lifetimes due to reaction with OH radical in thetroposphere below 300 mbar or the closest modellevel, calculated by the models to vary between 6.6and 10 yr (median 7.95 yr). Di�erences in thecalculated CO surface distributions can be partlyattributed to the surface losses of CO neglected inseveral models (see Table 3). Daniel and Solomon(1998) show that accurate assessment of thisquantity is important at a climatic time-scale.However, as shown in Table 3, this loss couldcontribute about 10% to the CO removal in theatmosphere, which is in the range of the uncer-tainties of CO sources.
4.2. Middle troposphere CO distributions
The calculated CO distributions in the middleand upper troposphere are of particular interest (i)because they a�ect the OH maximum in the middletropical troposphere and thus CH4 levels and (ii)they modify ozone concentrations in the uppertroposphere where ozone radiative impact maxi-mises. These CO distributions are even more dif-ferent between the models than the surfacedistributions since they strongly depend on thetransport of CO in the atmosphere by advection,di�usion and convection. In particular di�erencesin the convection parameterisation result in dif-
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 277
ferent patterns in the middle Figs. 5 and 6 andFigs. 7 and 8 and upper tropical troposphere (notshown).
In the middle troposphere at 500 mbar Figs. 5and 6 and Figs. 7 and 8 the detailed signature ofthe surface sources of CO is lost since transportzonally mixes CO. The sharp interhemisphericgradient of CO remains obvious in the free tro-posphere in January. This pattern is almost lost inJuly in most models both in the boundary layerand the free troposphere, due to the relativelyshort lifetime of CO in NH summer that prohibitse�cient CO transport far from its technologicalsources. Simultaneously in the SH, the CO mixingratio is more uniform (wintertime conditions).
A notable feature of most models is that theyrepresent the propagation of the CO biomassburning sources to the free troposphere by deepconvection, which is very intense in the tropics.However, this phenomenon is represented verydi�erently by the various models, re¯ecting thedi�erences in the model parameterisation of theconvection.
In general, models calculate an important de-crease in CO concentrations between the boun-dary layer and the middle troposphere, themagnitude of which depends on the season, thegeographical location and the model. Thus, theCO concentrations at 500 mbar are 0.2±0.8 timesthose calculated for the boundary layer with the
lowest ratios calculated over continental intensivesource areas. The reductions in CO between thesurface and the 500 mbar over oceans range from0.6 to 1.0 (no reduction). At some tropical loca-tions models calculate higher CO concentrationsin the middle troposphere than in the boundarylayer, re¯ecting strong vertical convective trans-port and possibly di�erences in boundary layersinks of OH.
4.3. Comparison of model results to observations
The observed monthly averages between dif-ferent years (Novelli et al., 1992) have been com-pared to the calculated seasonal variation ofsurface CO concentrations at ®ve surface stations:Barrow (71°N, 157°W), Mace Head (53°N, 10°W),Mauna Loa (20°N, 155°W), Samoa (14°S, 171°W)and Cape Grim (41°S, 145°E). Table 4 depicts theannual mean calculated and observed CO con-centrations at these ®ve stations.
To compare the seasonal trends in the modelsafter removing the di�erences in annual meansbetween the models shown in Table 4, in Figs. 9and 10 the seasonal variation of the normalised tothe annual mean CO concentrations as calculatedby the eleven models and as observed at the ®vesurface stations have been plotted. These normal-ised concentrations correspond to the ratio of thedi�erence between the monthly mean concentra-
Table 4
Calculated and observed annual mean CO concentrations (in ppbv) at ®ve surface stations
Model Barrow Mace Head Mauna Loa Samoa Cape Grim
IMAU3 117.5 114.2 66.0 54.0 59.8
IMAGES 114.4 127.4 76.1 50.3 62.9
HARVARD 121.5 130.8 89.5 58.8 62.2
UKMETO 101.0 118.3 73.9 57.8 66.0
ECHAM 153.6 162.3 92.2 68.1 79.9
CTMK 111.4 107.7 67.3 53.2 58.5
MATCH 149.1 222.8 77.7 59.8 71.5
MOZART 160.2 150.8 78.4 58.1 68.9
TM3 124.6 184.7 79.5 51.0 61.5
MOGUNTIA 135.3 161.8 90.4 54.5 69.2
UIO 131.6 130.2 73.7 60.9 74.1
Observations 148.3 133.1 96.3 62.7 53.4
(138.6±148.3) (128.8±137.4) (89.6±103.0) (57.6±67.7) (49.9±57.0)
278 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
tion of CO from its annual mean value divided bythe annual mean.
It is notable that the di�erences between modelsare larger than the inter-annual variability associ-
ated with the observations. Most models tend tounderestimate CO levels at Barrow (NH) (Table 4)but capture the seasonal trend within 50% (Fig. 9),which indirectly is an indication for OH levels. At
Fig. 9. Seasonal variation of the surface CO concentrations normalised to their annual mean: Comparison between model results and
observations at Barrow, Mace Head and Mauna Loa.
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 279
Mace Head, an NH coastal station, two modelslargely overestimate the CO levels whereas theremaining are within 30% of the observed values.More models have di�culties in capturing theobserved seasonal trend at this location (Fig. 9).At Mauna Loa station, more representative of thefree troposphere, all models underestimate COlevels by 4±30 ppbv (i.e. 6±30%, Table 4) depend-ing on the model and overestimate the CO sea-sonal amplitude in particular in January (Fig. 9).At Samoa (SH) most models (only one exception)underestimate the CO levels by 3±20% and they donot capture its seasonal cycle. On the contrary, atCape Grim (SH) they overestimate the CO con-centrations by 10±50% as well as the seasonaltrend (Fig. 9).
These patterns can be related to (i) the adoptedemissions (see Table 3) and emission distributions
in time and space, (ii) to the calculated di�erencesin OH radical concentrations between the NH andthe SH which are also re¯ected in the lifetime ofmethane (CH4). As a matter of fact, all modelscalculate the photochemical CH4 lifetime in thetroposphere up to 300 mbar or the closest modellevel, to be higher in the SH (median 9.2 yr) thanin the NH (median 7.2 yr). This implies a lowerconcentration of OH in the SH than in the NH.However, the opposite trend in OH levels is ex-pected from the lower CO and CH4 concentra-tions in the SH as well as from 14COmeasurements (Warneck, 1988; Brenninkmeijer etal., 1992) which are indirect indications for OHlevels. This could indicate that NH industrialemissions might be underestimated in the models,in particular those related to biofuel and wasteburning.
Fig. 10. Seasonal variation of the surface CO concentrations normalised to their annual mean: Comparison between model results and
observations at Samoa and Cape Grim.
280 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282
The overestimate of CO concentrations at CapeGrim may also result from the fact that the ob-servations depicted in this ®gure are clean sectordata. The model results contain all sectors andinclude air masses originating the Australian con-tinent. They are therefore expected to be higherthan the clean sector observations. Sector selectionis also performed at the Mace Head station data,where some models tend to overestimate CO levelsand the range of model results largely exceeds thatof the observations.
5. Conclusions
The intercomparison of eleven 3-D globalchemistry/transport models shows signi®cant dif-ferences between the models although all of themcapture the general patterns in the observed globaldistribution of CO. The comparison betweenmodel results and observations at selected stationsallows the evaluation of the model deviations fromthe observed annual mean CO concentrations atabout �50%, whereas di�erences between modelsare more important since models can deviate fromtheir annual mean values by up to 80%.
More accurate emission inventories of CO andother ozone precursors, measurements of theglobal 3-D distribution of tropospheric CO, andbetter assessment of the chemical processes deter-mining the oxidizing capacity of the troposphereand especially governing the OH levels could helpimproving model simulations and reducing relateduncertainties and model deviations from reality.
Unfortunately within the present exercise theimportance of emission inventories and chemicalprocesses cannot be separated from other contrib-uting factors like transport and deposition di�er-ences. In future intercomparison exercises, modelsshould preferably use the same emission inventor-ies as input, thereby ruling out di�erences betweeninventories as cause of di�erences between models.
Acknowledgements
We thank Elena Abilova from the University ofNorway for centralising and drawing the 2-D plots
of CO, Ra� Semercyan for compiling the stationdata in 1997. The constructive comments of ananonymous reviewer are kindly acknowledged. Weexpress gratitude to the European Union, theNASA and the PNCA/INSU of the CNRS forsupporting this exercise. MK expresses gratitudeto IDRIS for computing facilities.
References
Bernsten, T.K., Isaksen, I.S.A., 1997. A global three-dimen-
sional chemical transport model: model description and CO
and ozone results. J. Geophys. Res. 101, 21239±21280.
Brasseur, G.P., Hauglustaine, D.A., Walters, S., 1996. Chem-
ical compounds in the remote paci®c troposphere, compar-
ison between MLOPEX measurements and chemical
transport model calculations. J. Geophys. Res. 101,
14183±14795.
Brasseur, G.P., Hauglustaine, D.A., Walters, S., Rasch, P.J.,
Muller, J.-F., Granier, C., Tie, X.X., 1998. MOZART: a
global chemical transport model for ozone and related
chemical tracers, Part 1. Model description, J. Geophys.
Res. 103, 28265.
Brenninkmeijer, C.A., Manning, M.R., Lowe, D.C., Wallace,
G., Sparks, R.J., Volz-Thomas, A., 1992. Interhemispheric
asymmetry in OH abundance inferred from measurements
of atmospheric 14CO. Nature 356, 50±52.
Collins, W.J., Stevenson, D.S., Johnson, C.E, Derwent, R.G.,
1997. Tropospheric ozone in a global scale three-dimen-
sional Lagrangian model and its response to NOx emission
controls. J. Atmos. Chem. 26, 223±274.
Crutzen, P.J., 1994. An overview of atmospheric chemistry. In:
Topics in Atmospheric and Interstellar Physics and Chem-
istry and ERCA, Les Editions de Physique, Les Ulis,
France, pp. 63±89.
Daniel, J.S., Solomon, S., 1998. On the climate forcing of
carbon monoxide. J. Geophys. Res. 103, 13249±13260.
Fung, I., John, J., Lerner, J., Matthews, E., Prather, N., Steele,
L.P., Fraser, P.J., 1991. Three-dimensional model synthesis
of the global methane cycle. J. Geophys. Res. 96, 13033±
13065.
Hao, W.M., Liu, M.H., Crutzen, P.J., 1990. Estimates of
annual and regional release of CO2 and other trace gases
to the atmosphere from ®res in the tropics, based on
FAO statistics for the period 1975±1980. In: Goldammer
(Ed.), Fire in the Tropical Biota: Ecosystem Processes
and Global Challenges. Springer, New York, pp. 440±
460.
Hough, A.M., 1991. Development of a two-dimensional global
tropospheric model: Model chemistry. J. Geophys. Res. 96,
7325±7362.
Houweling, S., Dentener, F.J., Lelieveld, J., 1998. The impact
of non-methane hydrocarbon compounds on tropospheric
photochemistry. J. Geophys. Res. 103, 10673-10696.
M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282 281
Isaksen, I.S.A., Hov, O., 1987. Calculations of trends in the
tropospheric concentrations of O3, OH, CO, CH4 and NOx.
Tellus B 39, 271±285.
Jacob, D.J., Prather, M.J., Rasch, P.J., 1997. Evaluation and
intercomparison of global atmospheric transport models
using 222Rn and other short-lived tracers. J. Geophys. Res.
102, pp. 5953-5970.
Kasibhatla, P.S., II, H., Klonecki, A., Chameides, W.L., 1996.
The Three-dimensional view of large scale tropospheric
ozone distribution over the North Atlantic during summer.
J. Geophys. Res. 101, 29305±29316.
Krol, M., Weele, M., 1997. Implications of variations in
photodissociation rates for global tropospheric chemistry.
Atmos. Environ. 31, 1257±1273.
Lawrence, M.G., Crutzen, P.J., 1998. The impact of cloud
particle gravitational settling on soluble trace gas distribu-
tions. Tellus 50b, 263±289.
Lawrence, M.G., 1996. Photochemistry in the tropical paci®c
troposphere: studies with a global 3D chemistry-meteorol-
ogy model, Ph.D. thesis, Georgia Institute of Technology.
Lawrence, M.G., Chameides, W.L., Kasibhatla, P.S., Levy II,
H., Moxim, W., 1995. Lightning and atmospheric chemis-
try. In: Volland, H. (Ed.), Handbook of Atmospheric
Electrodynamics, The rate of Atmospheric NO production,
CRC Press, Boca Raton, Florida, vol. 1, pp. 189±202.
Muller, J.F., Brasseur, G.P., 1995. IMAGES: A three-dimen-
sional chemical transport model of the global troposphere.
J. Geophys. Res. 100, 16455±16490.
Novelli, P.C., Steele, L.P., Tans, P.P., 1992. Mixing ratios of
carbon monoxide in the troposphere. J. Geophys. Res. 97,
20731±20750.
Olson, J., Prather, M., Rasch, P., 1997. Results from the
intergovernmental panel of change photochemical model
intercomparison (photo comp.). J. Geophys. Res. 102,
5979±5991.
Poisson, N., 1997. Impact des hydrocarbures non m�ethaniques
sur la chimie troposph�erique. Ph. D. thesis, Universite Paris
7, France.
Poisson, N., Kanakidou, M., Crutzen, P.J., 1999. Impact of
non-methane hydrocarbons on tropospheric chemistry and
the oxidizing power of the global troposphere: 3-dimen-
sional modelling results. J. Atmos. Chem., submitted.
Prather, M.J., 1996. Time scales in atmospheric chemistry:
theory, GWPs for CH4 and CO, and runaway growth.
Geophys. Res. Lett. 23, 2597±2600.
Price, C., Rind, D., 1992. A simple lightning parameterisation
for calculating global lightning distributions. J. Geophys.
Res. 97, 9919±9933.
Roelofs, G.J., Lelieveld, J., 1995. Distribution and budget of O3
in the troposphere calculated with a chemistry general
circulation model. J. Geophys. Res. 100, 20983±20998.
Roelofs, G.J., Lelieveld, J., 1997. Model study of the in¯uence
of cross-tropopause O3 transports on tropospheric O3 levels.
Tellus 49B, 38±55.
Stordal, F., Derwent, F.G., Isaksen, I.S.A., Jacob, D.,
Kanakidou, M., Logan, J.A., Prather, M.J., 1995. Model
Simulations of Global Tropospheric Ozone, Ch. 7, Scienti®c
Assessment of Ozone Depletion/1994. United Nations
Environment Programme and World Meteorological Orga-
nisation, p. 33.
Wang, Y., Jacob, D.J., Logan, J.A., 1998a. Global simulation
of tropospheric O3±NOx-hydrocarbon 1. Model formula-
tion. J. Geophys. Res. 103, 10713±10726.
Wang, Y., Jacob, D.J., Logan, J.A., 1998b. Global simulation
of tropospheric O3±NOx-hydrocarbon 3. Origin of tropo-
spheric ozone and e�ects of non-methane hydrocarbons. J.
Geophys. Res. 103, 10757±10768.
Wang, Y., Logan, J.A, Jacob, D.J., 1998c. Global simulation of
tropospheric O3±NOx-hydrocarbon 2. Model evaluation
and global ozone budget. J. Geophys. Res. 103, 10727±
10756.
Warneck, P., 1998. Chemistry of the Natural Atmosphere.
Academic Press, INC.
Wauben, W.M.F., Fortuin, J.P.F., Velthoven, P.F.J., Kelder,
H., 1998. Comparison of modelled ozone distributions with
sonde and satellite observations. J. Geophys. Res. 103,
3511±3530.
Wauben, W.M.F., Velthoven, P.F.J., Kelder, H., 1997. A 3D
chemistry transport model study of changes in atmospheric
ozone due to aircraft NOx emissions. Atmos. Environ. 31,
1819±1836.
Yienger, J.J., Levy II, H., 1995. Empirical model of global soil-
biogenic NOx emissions. J. Geophys. Res. 100, 11447±
11464.
Maria Kanakidou is Assistant Professor of EnvironmentalChemistry at the Department of Chemistry of the University ofCrete, Heraklion, Greece. She has been co-convenor of theGlobal Integration Modeling (GIM) activity of IGAC of theIGPB since summer 1996 and a member of the Commission onAtmospheric Chemistry and Global Pollution (CACGP) of theIAMAS since summer 1998. In the past, she has worked as aresearch scientist at the Max±Planck-Institute for Chemistry inMainz, Germany, and at the CNRS in Gif-sur-Yvette, France.Her research concerns atmospheric chemistry and physics witha focus on tropospheric chemistry and climate changes due tohuman activities and on gas/particle interactions (http://ecpl.chemistry.uch.gr/cv/kanakidou.html).
282 M. Kanakidou et al. / Chemosphere: Global Change Science 1 (1999) 263±282