JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. D4, PAGES 7199-7218, APRIL 20, 1993
A Three-Dimensional Modeling Study of Trace Species in the Arctic Lower
Stratosphere During Winter 1989-1990
M.P. CHIPPERFIELD, 1 D. CARIOLLE, AND P. SIMON
Mdto-France, Centre National de Recherches Mtdorologiques, Toulouse, France
R. RAMAROSON
Offrice National d'Etudes et de Recherches Arospatiales, ChdtiIIon, France
D. J. LARY
Centre for Atmospheric Science, University of Cambridge, Cambridge, Engiand
A three-dimensional (3D) radiative-dynamical-chemical model has been developed and used to
study the evolution of trace gases in the Arctic lower stratosphere during winter 1989-1990. A
series of 10-day model integrations were performed throughout this period. The model includes a
comprehensive scheme of gas phase chemical reactions as well as a parameterization of heteroge-
neous reactions ocmtrring on polar stratospheric cloud (PSC) surfaces. An important element of
a 3D chemical model is the transport scheme. In this study the transport of chemical species is
achieved by a non diffusive method well suited to the preservation of sharp gradients. During the
winter studied temperattires were cold enough for the formation of both type I and type II polar
stratospheric clouds from early December to early February. Model simulations in late December
show that inside the polar vortex air is rapidly processed by polar stratospheric louds converting
HC1 and C1ONO2 to active chlorine. The possibility of ozone destruction depends strongly on
the amount of sunlight. In early February an average ozone loss of 15 ppbv (parts per billion by
volume) /day is predicted in PSC-processed air at 50 hPa, giving a column loss of just under 1
DU/day. This loss increases to 25 ppbv/day if PSCs persist until March with a column loss of
around 1.5 DU/day. The relatively small magnitude of the ozone loss predicted in the model,
compared to the variability of ozone induced by dynamics, highlights the problems in identifying
the signatLre of chemical ozone loss in the Arctic. In furtire years significant ozone depletion could
occur if PSCs persist until late March. The efficiency of the catalytic cycles responsible for the
ozone loss has been analyzed as a function of latitude, altitude and time. In general, the cycle
involving C]O + C10 is the dominant loss mechanism in the polar h)wer stratosphere. Cycles
involving BrO can make a relatively large contribution early in the season and when the levels of
C10 are low. The cycle initiated by C10 + O destroys ozone at altitudes above 30 hPa but the
loss is compensated, to some extent, by in situ ozone production. The results for trace species are
validated, where possible, by comparison with the available measurements, although the sparse
nature of the observations does not effectively constrain the model.
1. ]NTRODUCTION
It is now well established that the springtime ozone de-
pletion observed over Antarctica is caused by reactions of
the surface of polar stratospheric louds (PSCs) altering the
normal gas-phase partitioning of odd nitrogen (NOv) and
odd chlorine (ClOy) species. This enables catalytic cycles
involving C10 and BrO radicals to efficiently destroy ozone
[see, for example, Solomon, 1990]. Of prime concern now is
the extent to which similar processes could occur over the
Arctic during winter and spring. The north polar vortex is
more dynamically disturbed than its southern counterpart
giving warmer temperatures and therefore less opportunity
for PSCs to form. Also, the northern hemisphere vortex
breaks down earlier, preventing ozone depletion from oc-
curring significantly into the spring season. However, the
azonal nature of the circulation permits air which has expe-
rienced cold temperatures at the center or edge of the vortex
1Now at Centre for Attnospheric Science, University of Cam-
bridge, U.K.
Copyright 1993 by the American Geophysical Union.
Paper number 92JD02977.
0148-0227/93/92 JD-029 ??$05.00
to experience significant exposure to sunlight.
The interest in the chemistry of the Arctic stratosphere
has prompted many measurement campaigns. In January
and February 1989 the Airborne Arctic Stratospheric Ex-
periment (AASE) employed two aircraft to make measure-
ments of several trace species (e.g., see Geophysical Research
Letters volume 17, number 4, March supplement 1990). The
following year the CHEOPS III (Chemistry of Ozone in the
Polar Stratosphere III) campaign [Pommereau and Schmidt,
1991] made balloon-borne and ground-based observations
from Scandinavia and Greenland. These campaigns showed
that the chemistry of the Arctic lower stratosphere was per-
turbed in a similar way to the Antarctic with, for example,
high levels of C10. Thus the north polar vortex was consid-
ered as being \"primed\" for ozone destruction and, although
no large-scale ozone loss was observed, localized depletion
of around 0.4% per day has been reported [e.g., Schocbcrl et
al., 1990, Salawitch et al., 1990].
Kaye et al. [1991] and Douglass et al. [1991] used a three-
dimensional (3D) transport model, with a parameterization
of the chemistry of H C1, to investigate chemical process-
ing by PSCs. They studied the winters of 1979 and 1989
using assimilated meteorological data and concluded that
the transport of processed air, as signalled by perturbed
HC1, to mid-latitudes was limited. Granier and Brasseur
7199

7200 CHIPPERFIELD El' AL.: THREE-DIMENSIONAL MODELINO STUDY
[1991] used a mechanistic 3D model with a fairly detailed
chemistry scheme to investigate the potential for large-scale
ozone loss over the Arctic. They applied a climatological
forcing to their model and investigated the effects of inter-
hemispheric differences in dynamics on the potential for an
ozone loss over the north pole. In this study, rather than
using climatological winds, we have used a 3D radiative-
dynamical-chemical model to investigate trace species distri-
butions and chemical ozone destruction by performing \"real
case\" studies of the Arctic lower stratosphere during win-
ter 1989-1990. We have used a model with a more detailed
chemistry scheme and where possible we have validated it
against the available observations. We have attempted to
make the model as well suited as possible to the short-
term (10-day) integrations by paying particular attention to
the initialization of the model meteorological and chemical
fields.
During the northern winter of 1989/1.990 the midwinter
period was extremely cold with temperatures in the polar
lower stratosphere below the seasonal average [Naujokat et
al., 1990]. There was a minor warming in early December
but it did not have a significant effect on temperatures in-
side the polar vortex. Temperatures remained cold through-
out January and into February followed by a stra.tospheric
warming which did not break down the polar vortex. The
final warming was late, with the temperature gradient fi-
nally reversing in mid April. Figure 1 plots the minimum
temperature in the ECMWF (European Centre for Medium
Range Weather Forecasts) analyses between 50øN and the
pole at the 30 hPa and 50 hPa levels from November 1989
to April 1990. Also indicated are the temperatures below
which type I and type II PSCs are believed to form. The
Figure shows that type I PSCs would have begun to form in
mid December at these two altitudes and persisted through-
out the cold midwinter period until early February. Type II
PSCs could have formed in late January and early February.
After the pronounced minor warming around February 10
the possibility of PSC formation is greatly reduced. Tem-
peratures at 30 hPa remain at least 5 K above the type I
PSC formation temperature. At lower levels (50 hPa and
below) type I PSCs may have briefly formed in mid Febru-
ary and mid March. Whilst the ECMWF analyses capture
the large-scale features of the lower stra.tospheric tempera-
ture field they often miss small, localized temperature vari-
ations. Notably, orographic forcing and adiabatic cooling of
air passing over the Scandinavian mountain range can cause
temperatures cold enough for PSCs which are not revealed
in the ECMWF analyses.
During a series of balloon flights from January 8 to the
February 8 1990, Ho/mann and Deshler[1991] measured the
ozone profile above Kiruna (68øN, 21øE). From their data,
they inferred that chemical ozone destruction had occurred
at around 22 km during this period. First, from the Jan-
uary 12 to 19 January an \"episodic\" loss of 1 ppmv (parts
per million by volume) (140 ppbv (parts per billion by vol-
ume) /day) occurred between 22 and 26 km followed by a
gradual reduction of around 20 ppbv/day throughout late
January and early February. Similarly Koike et al. [1991]
used balloon-borne measurements of ozone from Kiruna to
estimate the magnitude of chemical depletion observed in
January and Febr,uary 1990. They concluded that ozone
destruction of 0.8% per day occurred on the 525 K isen-
tropic surface betweelt January 18 and February 4 inside
the vortex, which coincided with temperatures cold enough
for typ.e  IPSC formation. ! the latter part of this period,
after January 26 the estimated loss rate was 1.5% per day
(48 ppb¾/day).
Le/Jvre et al. [1991] and Riishojgaard et al. [1992] used
a general circulation model (GCM) with a parameterization
of the gas phase chemistry bfozone to study the type II
220
25-
210
205
I I
t'O 200 '\"..
_.
E
.
190
i
180  ,
-60
Minimum Temperature at 30 hPa
I I I I
-40 -20 0 20 40 60 80 1 O0
Day (after January 1 1990)
120
Fig. 1 a. Minimum temperature (K) in the ECMWF analysis at 30 hPa between 50øN and 90øN from November
1989 to April 1990. Also indicated are the temperatures below which type I and type II PSCs form.

CHIPPBRFIBLD El' AL.: THREE-DIMENSIONAL MODBLINO STUDY 7201
220
215
210 -
' 205-
(1;i 200-
E
195
190
185
180 [
-60
Minimum Temperature at 50 hPa
-40 -20 0 20 40 60 80 100
Day (after January 1 1990)
Fi$. lb. Sae as Fitre ta but œor 50 hF'a.
120
PSC event of early February 1990 using TOMS (Total Ozone
Mapping Spectrometer) data to derive the initial model Oa
field. By employing tracers to indicate the amount of time
air had spent in sunlight and in the presence of type II PSCs,
and comparing the evolution of the model column ozone
field with TOMS data, Le/vre et al. [1991] inferred that
substantial ozone loss of 50 DU in 3 days (equivalent o a
40% loss at 18 km) occurred in air which had been processed
by the type 1I PSC and had spent 16 hours in sunlight.
In our study we have used a three-dimensional chemical
transport model (CTM) which uses winds derived from a
GCM. The CTM has a detailed chemical scheme involving
the O, HO, NOu, C10 u and BrO u families and a param-
eterization of heterogeneous reactions on PSCs. The trans-
port of tracers is achieved using the second order moments
scheme of Prather [1986] which is well suited to the preser-
vation of strong species gradients. We have performed a
series of 10-day experiments spread throughout the winter
of 1989-1990 to study the evolution of ozone and other trace
gases. The model results are validated, where possible, by
comparison with the available measurements, and the mag-
nitude of ozone loss predicted has been analysed in terms of
the chemical cycles responsible. Our model experiments for
early February 1990 allow us to investigate the conclusions
of Le/gre et al. [1991] further, using a chemical 3D model,
by examining the extent to which PSC processing may have
lead to chemical ozone destruction during this period. The
results of our model simulations are presented below. First,
section 2 describes the Emeraude general circulation model
and the chemical transport model, including the treatment
of PSCs, which have been used. Section 3 describes the
initialization of the model with the meteorological analysis
and two-dimensional (2D) model chemical fields transformed
into the coordinates of potential temperature and potential
vorticity. The results of the experiments are presented in
section 4, and the conclusions are summarized in section 5.
2. TH MODS
2.1. The \"Emeraude\" GCM
In this study we have used the tropospheric-stratospheric
version of the 'Emeraude\" GCM described by Cariolle et
al. [1990]. The model extends from the ground to about 80
km using a hybrid sigma-pressure coordinate scheme. In the
lower stratosphere the vertical resolution is about 1.8 kin.
The model contains a full description of the main physical
processes such as radiative transfer and the hydrological cy-
cle. This spectral model has been used in the horizontal res-
olution of T21, i.e., allowing a maximum total wavenumber
of 21 in the horizontal. The Gaussian grid corresponding to
this truncation has a resolution of 5.6 ø in longitude and lat-
itude. The model uses a leapfrog integration scheme with a
30-min time step. An integration of the GCM is performed
to generate wind and temperature fields which are stored
every 6 hours and used to force the CTM.
2.2. Chemical Transport Model
The chemical transport model uses the same grid as the
GCM which, for the experiments described here, is the
Gaussian grid corresponding to the T21 spectral resolution.
The method by which tracers are transported is critically
important in coupled chemical-dynamical models. This is
especially true for studies of the polar lower stratosphere
where localized PSC processing can lead to strong inhomo-
geneities in a tracer's distribution. In our CTM the chemical
tracers are advected using the transport scheme of Prather
[1986] in three dimensions which conserves the second-order
moments of the tracer distribution. This scheme has been
used, in preference to the spectral transport scheme of the
GCM, as it is less diffusive and better able to represent the
strong gradients in species distribution that can be found,
for example, at the edge of the polar vortex. The use of

7202 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY
the Prather scheme has resulted in a large improvement in
the model compared to a previous version which used the
spectral scheme.
The photochemical package which has been coupled to
the CTM is based on a photochemical box model, devel-
oped by Rarnaroson [1989] and Rarnaroson et al. [1990,
1992]. The model uses the semi-implicit symmetric (SIS)
method to integrate the chemical continuity equations. Ef-
fectively, this box model is called at every model grid point
and the chemical changes calculated are added to the mean
tracer distribution (zeroth-order moment) of the Prather
scheme in each box, splitting the effects of chemistry and dy-
namics (the first- and second-order moments of the tracer
distribution are not modified by the chemistry). For this
study the photochemical model of Ramaroson et al. [1990]
has been updated as follows. The chemical species O, 03,
H2Oa OC10, ClaOa Br, BrO, BrONOaand BrC1 have been
added to make the model appropriate for the study of po-
lar stratospheric chemistry. Table I lists the transported
chemical species in the model and the HOx species which
are assumed to be in steady state. Note that in the model
the chemical families NOr, C10 and BrO have no sources
or sinks, which is a satisfactory approximation to make in
the lower stratosphere for simulations of 10 days or so. The
mixing ratio of HaO is obtained from the water vapor tracer
of the GCM. In addition, fixed zonal mean fields of CH4 and
CO, taken from the 2D model of Cariolle and Brard [1984],
are used. The distribution of CH4 is necessary to determine
the recovery of C10 species after heterogeneous process-
ing. The model uses the family approach for certain short-
lived species (e.g., O + 03, G1 + G10 + Gla2, Br + BrO,
NO q- NO2 q- NO3). These species rapidly establish pho-
tochemical equilibrium during daylight and the use of these
families does not prevent the partitioning of species during
darkness. This has the advantage that the concentration of
each species is calculated at every location, which is impor-
tant when considering localized sources and sinks of species
due to reactions on PSC surfaces. When studying the po-
lar lower stratosphere, the calculation of photolysis rates is
an important consideration. For this study we have used
the \"look-up table\" scheme of Lary and Pyle [1991] which
is both computationally efficient and able to calculate pho-
tolysis rates for zenith angles up to 96 ø. This scheme does
not use the model calculated O3 field in the calculation of
the optical depth but instead uses a standard mid-latitude
O3 profile. The photochemical data is taken from DeMote
et al. [1990]. The absorption cross sections for BrC1 were
obtained from R. A. Cox (personal communication, 1988).
For the photolysis of HNO3 the temperature dependent cross
sections of Rattigan et al. [1992] were used. The time step
used in the GTM is 30 min for the dynamics and 15 rain for
the chemistry. Although the chemistry scheme is stable for
longer time steps a short time step was chosen so that the
diurnal cycle of species could be adequately resolved.
In addition to these improvements in the chemical scheme
a treatment of heterogeneous processes, occurring on the
surface of polar stratospheric clouds, has been included in
the model. The scheme does not contain any detailed mi-
crophysics but it does distinguish between type I and type
II clouds. For type I PSCs the model temperature and mix-
ing ratios of H20 and HNO3 are used, together with the
algorithm of Hanson and Mauersberger [1988a], to predict
when these PSCs are thermodynamically possible and the
resultant equilibrium gas phase mixing ratio of HNO3. The
condensed particles of HNO3.3H20 (nitric acid trihydrate -
NAT) are assumed to have a radius of 1 ttm, which is used
in the calculation of the available surface area. For type II
PSCs the model temperature field and water vapor is used in
conjunction with the Tetens equation [Murray, 1967], which
describes the saturation vapor pressure of water over ice, to
predict their occurrence. Similarly, when type II PSCs are
thermodynamically possible, the amount of ice condensed is
assumed to be in particles of radius 10 tim. When formed,
these type II particles are sedimented with a speed of 1.5
km/day, consistent with a size of 10 tim. The sedimented
PSC particles are deposited in the model level below. HC1
is incorporated into the PSCs in the mole fractions given by
Hanson and Mauersberger [1988b]. When the PSCs evap-
orate the condensed species are returned to the gas phase.
Once PSCs have been formed in the model, the following
five reactions are assumed to occur on the particle surfaces:
(R1) N2Os(g) + H20(s) - 2HNO3(s)
(R2) N2s(g) + HCI(s) - C1NO2(g)+ HNO3(s)
(R3) C1ONO2(g) q- adO(s)  HOCl(g) q- HNO3(s)
(R4) C1ONOa(g) + HCI(s) - Cla(g) + HNO3(s)
(RS) HOCI(g) q- HCI(s) - Cla(g) q- H20(s)
The reaction probabilities (7) for these reactions are taken
from experimental data and the values used are given in Ta-
ble 2 for the two types of PSC (see World Meteorological
Organisation/UNEP [1991] for a discussion of vaue).
Reaction (R4) is believed to involve the two step process of
reaction (R3) followed by reaction (RS) tAbbutt et al., 1992].
Therefore for reaction (RS) the same values of 7 were em-
ployed as for reaction (R4). The rate of chemical reaction
is calculated by using the measured reaction probabilities
listed in Table 2 along with the calculated PSC surface area
and the mean kinetic speed of the gaseous molecules [see,
e.g., Rodriguez et al., 1989]. These reactions are treated as
first-order loss reactions for the gaseous molecule unless the
concentration of condensed H C1 would limit this rate. Note
that in order to maintain the computational efficiency of
the model th hoine products of reactions (R2), (R4) and
(R5) are given in terms of C1 and NO2. In the stratosphere
any C12 and C1NO2formed will be rapidly photolyzed in
the presence of sunlight. In the dark the C1 produced in
the model by the heterogeneous reactions will be instan-
taneously converted to C10 and CI202 within the model
C1Ox family. With the adopted parameterization type II
PSCs, when they occur, will provide faster heterogeneous
TABLE 1. Chemical Species Contained in the Model
Transported
(= + o(P) +
NOx (=NO + NO2 + NO3),NeO, HNOs,HO2NO2,
C10 (=C1 + C10 + 2CleOe),C1ONO2,HC1,HOC1, OC10,
BrOx (=Br + BrO),BrONO2,BrC1
Not transported
HO (= H + OH + HOa)

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY 7203
TABLE 2. Reaction Probabilities for Heterogeneous Reactions
Reaction Probabihty
Reaction Type I Type II
(R1) 0.0006 0.03
(R2) 0.003 0.03
(R3) 0.006 0.3
(R4) 0.3 0.3
(RS) 0.3 0.3
rates than type I PSCs. First, the 7 values for the type II
particles are equal to or larger than those for type I parti-
cles. Second, type II PSCs will provide a larger surface area
per unit volume with the particle sizes adopted. In addi-
tion, type II PSCs will also perturb the atmosphere by the
permanent removal of NOy by sedimentation.
We have attempted to make the paramctcrization of PSCs
and the associated heterogeneous reactions as realistic as
possible without using a detailed microphysical model. How-
ever, there still exist a number of uncertainties. These in-
clude the variation of 7 as a function of particle composition,
the size distribution of particles expected under Arctic cool-
ing conditions and the onset temperatures of the formation
of type I PSCs. I-Ietcrogeneous reactions occurring on the
background sulphate aerosol layer are not included in this
study. (Note that by using LIMS (Limb Infra red Monitor
of the Stratosphere) data for the initialization of HNO3 we
have implicitly accounted for the effects of reaction (R1) on
the background aerosol.)
3. INITIALIZATION
For the experiments performed here the GCM was initial-
ized from the meteorological analysis of the ECMWF (Eu-
ropean Centre for Medium Range Weather Forecasts) for
the appropriate day. This analysis gives the initial fields of
the temperature, winds and water vapor below 10 hPa. The
initial conditions for the nine model levels located above 10
hPa wcrc taken from climatology.
The species in the CTM were initialized from fields taken
from an updated version of the 2D model of Harwood and
Pyle [1975]. However, this 2D model in common with others
has a few notable shortcomings at high latitudes. First, dur-
ing the winter at northern high latitudes 2D models which
contain only gas phase chemistry underestimate the abun-
dance of HNO3 when compared to LIMS data, probably
due to the omission of reaction (R1) above [Austin et al.,
1986], although problems at mid-latitudes may still remain
[Rood ½t al., 1990]. Therefore the 2D model field of HNO3
was constrained to agree with the LIMS (Limb Infra-red
Monitor of the Stratosphere) monthly mean data for Jan-
uary 1979 by including the effects of reaction (R1). Second,
2D models appear to underestimate the degree of descent
observed over the poles. This implies, for example, that
these models will underestimate the abundance of inorganic
chlorine (and overestimate, e.g., CH4) in the polar lower
stratosphere. This is an important factor in determining
the potential for ozone depletion at high latitudes and ac-
cordingly the initial 2D model field of ClOy was constrained
to agree with the estimates of the total inorganic chlorine
at 44øN and 67øN by Schmidt et al. [1991] based on their
measurements of long-lived tracers. The 2D model field of
inorganic bromine (BrOy) was constrained in a similar way.
In the model runs presented here the abundance of ClOy
and BrOy was 3.2 ppbv and 12 pptv (parts per trillion by
volume) respectively, in the upper stratosphere.
For short integrations of a 3D model the quality of the
simulation depends critically on the initialization of the
chemical fields. Traditionally, zonal mean fields from 2D
models have been used. However, during the northern win-
ter the atmosphere can be highly non zonal. Therefore in
this study we have attempted to initialize the chemical fields
in as realistic a way as possible by introducing the azonality
of polar vortex into the species distributions. This we have
done by initializing the fields in the north polar lower strato-
sphere using the coordinates of potential vorticity (PV) and
potential temperature (0). Douglass et al. [1990] described
the construction of global 3D fields in these coordinates us-
ing 2D model data. A similar procedure was adopted here.
The initial 3D model chemical fields were zonal mean
fields interpolated directly from the 2D model fields except
polewards of 30øN between 200 hPa and 10 hPa. In this
region the model fields of the longer-lived and more abun-
dant species were initialized using potential vorticity and
potential temperature. The zonal mean fields of potential
temperature and potential vorticity were calculated from the
2D model temperature field and these wcrc used to map the
constituent mixing ratios onto the 3D model grid using the
PV and 0 at each grid point. Thus the model fields were
constrained to follow the morphology of the polar vortex,
which, during this period, was highly non zonal. This is
illustrated in Figure 2 which shows PV and HNO3 on the
500 K iscntropic surface on February 4th, the initial day of
model run C (below). The PV map (Figure 2a) shows the
distorted nature of the polar vortex, with it displaced off of
the pole towards northern Europe. The HINO3 field (Figure
2b) mimics the distribution of PV. The HNO3 distribution
has weak horizontal gradients at mid-latitudes and in the
center of the vortex separated by strong gradients in mix-
EMERAUDE POTENTIAL VORTICITY 500 K
Exp. D 90 2 4 12H
Fig. 2a. Ertel's potential vorticity (x106 K m 2 s -1 kg -1) on
the 500 K isentropic surface at 1200 UT on February 4. Contour
interval is 10x106 K m 2 s -1 kg -a.

7204 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY
CTM HN03 (ppbv) 500 K
Exp. D 90 2 4 12H
Fig. 2b. Mixing ratio of HNO3 (ppbv) on the 500 K isentropic
surface at 1200 UT on February 4 from run D. Contour interval
is 0.5 ppbv.
ing ratio around the edge of the vortex. The distribution of
HNO3 around the 60 ø latitude circle illustrates the azonal
nature of the field, with values ranging from less than 5
ppbv to more than 10 ppbv. Note that the HNO field was
constructed by using a direct correlation between the zonal
mean PV calculated from the 2D model and the PV calcu-
lated in the 3D model. It is to be expected that the zonal
mean PV from the 2D model will not show values as large
as those observed at the center of the polar vortex. This
results in the large, featureless area inside the polar vortex.
In experiments C, D, E and F (see below) TOMS data was
used to initialize the model ozone field using the method of
Riishojgaard et al. [1992]. Finally, the shorter-lived chem-
ical species in the model were \"balanced\" by performing
a 2-day integration of the CTM without transport and a
constant temperature field. This was to ensure that at the
start of the model run proper these shorter-lived species had
been brought into equilibrium with the local temperature
and photolysis fields. During this 2-day integration the het-
erogeneous reactions described above were included where
appropriate. Thus at the start of the experiment itself the
air in the presence of the PSCs had already been perturbed.
In addition, because LIMS HNO data was used in the ini-
tialization procedure the partitioning of NO v species had
been perturbed all over the polar region from that expected
on the basis of gas phase chemistry. For some experiments
later in the winter, as described below, some preprocessing
of chlorine species was assumed.
4. RESULTS
A series of 10-day model integrations were performed over
the winter period. Runs A and B both started on the De-
cember 25 1989, shortly after the temperatures in the vortex
were cold enough to permit the formation of PSCs (Figure
1). Run A and run B differ only in that run A did not
contain heterogeneous chemistry while run B did. Two fur-
ther runs (C and D) were initialized on the February 4 1990.
In run D we assumed that prior to this date air in the po-
lar vortex had been processed by PSCs converting HC1 and
C1ONO2 to active chlorine. During early February the polar
vortex was displaced over Scandinavia with very cold tem-
peratures (cold enough for the formation of type II PSCs,
Figure 1) in the lower stratosphere. The position of the
vortex enabled air which had been processed by PSCs to
undergo exposure to sunlight. As shown in Figure 1, early
February provided the latest opportunity for significant PSC
activity. However, to examine the potential for ozone de-
pletion in winters where the temperatures remain cold until
later in the year, two experiments (E and F) were performed
using the same initial conditions and dynamical forcing as
experiments C and D but with the solar position advanced
by one month giving photolysis conditions appropriate for
early March. The model runs performed are summarized in
Table 3.
.1. Meteorological Forecast and PSCs
The simplified treatment of PSCs contained in the model
assumes that PSCs occur whenever they are thermodynami-
cally possible. Therefore as there is no barrier to nucleation,
the predicted area of PSCs will represent an upper limit.
However, this will be offset slightly because, as noted above,
the temperature forecast of the model differs from observa-
tions by a few degrees. Notably, the model does not produce
the very cold temperatures in or near the polar vortex caused
by orographic forcing and adiabatic ascent. Figure 3 shows
the predicted occurrence of PSCs at 50 hPa on February
5th from experiment D. At this altitude the extent of the
type IPSC approximately follows the 194 K temperature
contour and the type II PSC the 190 K contour. Figure 4 is
a latitude cross section at 33.75øE showing the occurrence
of type I and type II PSCs in the stratosphere (ice satura-
tion also occurs in the troposphere but is not shown). Type
I PSCs occur polewards of 60øN between 20 and 100 hPa.
(Interestingly, type I PSCs are also predicted at the equator
although, in the model, the heterogeneous chemical reac-
tions are not activated there.) Type II PSCs occur between
60ø-80øN and 70-30 hPa.
In our experiments the GCM, whose winds are used to
force the CTM, is used in the resolution of T21. In this low
resolution the GCM forecast will diverge from reality and
cannot be expected to completely capture the observed syn-
optic forcing. For example, in the experiments initialized on
February 4 the GCM does not faithfully reproduce the pro-
nounced minor warming which occurred around February
8.
TABLE 3. Model Runs Performed
Heterogeneous
Run Start Date Chemistry Comments
A Dec. 25 1989 no
B Dec. 25 1989 yes
C Feb. 4 1990 no
D Feb. 4 1990 yes
E Feb. 4 1990 no
F Feb. 4 1990 yes
Solar position
early March
Solar position
early March

CHIPPERFIELD ET AL.' THREE-DIMENSIONAL MODELING STUDY 7205
CTM PSC OCCURRENCE 50 hPa
EXP. D 90 2 5 12H
Fig. 3. The occurrence of type I (hatched shading) and type II
(solid shading) PSCs at 50 hPa at 1200 UT on February 5 from
run
4.2. Chemical Species
Results are presented here for trace chemical species.
Where possible the model values are compared to obser-
vations made during the CHEOPS III campaign and also
measurements from the AASE mission which will be repre-
sentative of the chemistry of the Arctic lower stratosphere.
The measurements serve to validate the model but are gen-
erally too sparse to effectively constrain its behavior.
Nitrogen species. Pommereau et al. [1991] measured the
vertical column of NOa with UV-visible spectrometers at
two sites during January and February 1990. One spec-
trometer was located at Sondre Stromfjord (67øN, 51øW)
and the other at Kiruna (68øN, 21øE). At Sondre Stromfjord
the observed twilight vertical column of NOa in January and
early February was around lx101 molecules cm - with the
morning and evening measurements yielding similar values.
During early February, coinciding with a minor warming,
the measured vertical column was 1.6x10 s molecules cm -
in the evening and 0.6 molecules cm - in the morning. At
Kitnrta the measured NO2 column in January and Febru-
ary showed a baseline of around 1.4x10 s molecules cm -
superimposed on which were numerous episodes of tropo-
spheric pollution leading sporadically to a much larger col-
umn. Adrian et al. [1991], using infrared measurements,
gave an upper limit to the vertical column of NO2 over Es-
range of 1.2x10 s molecules cm -. Figure 5a shows the NOa
column (above 300 hPa) from run D at 1200 UT on February
6th. At 30øN the diurnal variation of NO2 can be seen with
an increase in sunset followed by a slow decay during the
night and a sharp decrease at sunrise. There is a large area,
covering Scandinavia, the northern Atlantic and Greenland
where the N O2 column is less than lx10 s molecules cm -2
At 50 hPa (Figure. 5b) there are regions, north of 60øN be-
tween 30øE aqd 120øE, where the abundance of NO2 is ef-
fectively zero due to the heterogeneous conversion of N O
to HNO3. The use of LIMS HNO3 data to constrain the
initial model NO s fields results in the suppression of NO2
values throughout the polar region. The presence of PSCs
further reduces the NO2 over northern Scandinavia.
7'10\"
1.10 o-
2* 10 ø
4'10 ø
7* 10 ø
 1'10'
 2'10'
03 4'10'
I:L 7'10'
1.10 
2* 10 
4'10 
7'10  1'10 
CTM PSC OCCURRENCE
I I I I I I
-20 -10 0 10 20 30 40 50
Latitude (degrees)
60 70 80 90
Fig. 4. The predicted occurrence of type I PSCs (ver[ical shading) and type II PSCs (horizontal shading) at
33.75øE at 1200 UT on Februm'y 5 from run D.

7206 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY
CTM Column NO2 (10E15 molec.cm-2) CTM NO2 (ppbv) 50 hPc
Adrian et al. [1991] also measured the column abundance
of HNO3. In January and February 1990, above Kiruna
the HNO3 column abundance was large, being 2.9x10 6
molecules cm -2. This is consistent with the results from
the AASE and in contrast to observations in the Antarctic.
The column HNO3 abundance in the model (not shown) is
consistent with these observations. The column abundance
is around 2.5x10 6 throughout the polar region. The effect of
using LIMS data to initialize the HNO3 is to ensure that the
majority of NO s is contained in HNO3 in the lower strato-
sphere. The mixing ratio of HNO3 at 50 hPa on February 6
from run D is shown in Figure 6 (the quantity plotted is the
total HNO3 comprising both the gas phase and that con-
densed in the PSCs). The high values of HNO3 of around
9-10 ppbv are located inside the polar vortex. However, co-
incident with the type II PSC (Figure 4 above) there is a
local minimum with an equivalent gas phase mixing ratio
of 6 ppbv. This is caused by the sedimentation of the large
type II particles to lower altitudes. In these short, 10-day in-
tegrations the degree of denitrification will not significantly
affect the magnitude of the ozone destruction. Essentially
all of the odd nitrogen is in the form of HNO3, whose pho-
tolysis rate is slow enough (especially when calculated using
the temperature dependent cross sections of Rattigan et al.
[1992]) to prevent the release of NO2, thereby maintaining
the levels of active chlorine. The amount of denitrification
will, of course, affect the long-term recovery of an air mass
[e.g. see Prather and Jaffe, 1990].
Chlorine species. The processing of chlorine species in
early winter has been investigated by a comparison of ex-
periments A and B. Figure 7a shows the difference in HC1
between these two runs on the 500 K surface on the Jan-
uary 4 1990, 10 days into the model integration. Figure
7b shows PV on the 500 K isentropic surface on this day.
Air inside the vortex, as indicated by the strong gradients
CTM HN05 (ppbv) 50 hPa
......... '\"'. 6 i .' ,,,,/.::;-'-:i.!.,,.'. ..... . .. V .
-/.., '.. ,\"<'\" ,, :..,.<.. -. , ..... ?,
.-<.../:\"- ,-,-:,. ... ,
..
EXP. D 90 2 6 12H
Fig. 6. Mixing ratio of total HNO3 (ppbv) at 50 hpa at 1200 UT
on February 6 from run D. Contonr interva.] is 1 ppbv.
of PV, has been processed by PSCs resulting in the het-
erogeneous conversion of HC1 to active chlorine (C10). At
the center of the vortex effectively all of the H C1 has been
converted in run B. Note the way that the HCl-poor air
is confined to the polar regions and the strong gradients
at the edge of the vortex, even with the comparatively low
model resolution of T21. This illustrates the non diffusive

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELINO STUDY 7207
CTM DIFFERENCE HCI (ppbv) 500K EMERAUDE POTENTIAL VORTICITY 500 K
EXP B - A 90 1 4 12.H EXP. A 90 1 4 11H
Fig. ?a. Difference in the HCI mixing ra[io (ppbv) be[ween run Fig. ?b. ErCel's po[en[ial vor[ici[y (x106 K m 2 s -z kg -z) on [he
B and run A on [he 500 K surface a[ 1200 UT on January 4. 500 K isen[ropic surface a[ 1200 UT on January 4 from run A.
Con[our in[erval is 0.2 ppbv. Con[our in[erval is 5x106 K m 2 s -1 kg -1.
CTI,,4 CION02 (ppbv) 500 K
 -.
'., . _;,,'.. <-..,?. .... ..... ':'x: .... ' '
........ ;;, .............. \"', ........ i'jj . j\"'}'\"..../:\"' \"/:'\"J ..... -?\": .....
.' \"- C / ' .. .,::/.. --.
, ;\" ..; _.j
EXP. B 90 1 4 12H
Fig. ?c. Mixing ra[io of C1ONO2 from run B on [he 500K surface
a[ 1200 UT on January 4.
CTM ClOx (ppbv) 500 K
,/
, ' : ..... .i ... 2 ';:?- .....
, ,_., .::,,'
,,' i\" .. 'i. ,;-:.'F 'x::,...,'
..... /:'\"-i.' i .... . ...... ,,. /. ;, ..., ß ... ,,,, \"' ';. . ..... -': ,, i .,- ':::''-' ........... '\" . ,
EXP. B 90 1 4 12H
Fig. ?'d. Mixing ra[io oœ ClO: from run B on [he 500 K sur_½a, ce
a[ 1200 UT on January 4.
nature of the Prather transport scheme. In earlier experi-
ments we used the spectral scheme of the GCM for tracer
advection. The results were far more diffusive with changes
to species at high latitudes being rapidly %ransported\" to
lower latitudes. When this lead to chemical reactions be-
tween species the model results quickly became unrealistic.
For example, the Gibb's phenomenon, associated with the
spectral truncation, caused polar air high in CIO to react
wih mid-latitude Mr containing NO2 resulting in the for-
mation of C1ONO2. Whilst the performance of the spectral
scheme would improve at higher resolutions, the computa-
tional cost of including chemistry in a 3D model often neces-
sitates the use of a coarse resolution. In such low resolutions
the shortcomings of a transport scheme can be more evident.

7208 CHIPPERFIELD ET AL.' THREE-DIMENSIONAL MODELING STUDY
The properties of various transport schemes have been well
investigated in several 1D passive tracer experiments (e.g.,
see Muller [1992 and references therein]). However, when
chemical interactions between tracers are considered, the
problems are compounded and the results from a scheme
which is too diffusive can rapidly become unrealistic.
Figure 7c plots C1ONO2 from run B. The minimum at
the center of the vortex is partly due to loss by heteroge-
neous reactions on PSCs. This results in a ring of C1ONO2
around the pole. Note that in run B mixing at the edge
of the vortex has not lead to a large increase in C1ONO2
compared to run A. Mixing has contributed to an increase
of C1ONOe of 0.2 ppbv compared to run A at 500 K. At 600
K (not shown) the effect of mixing is larger with an increase
in C1ONOe of 0.6 ppbv. Figure 7d shows the mixing ratio
of C10: (=C10 + 2CleOe) at 500 K from run B. Inside the
polar vortex the mixing ratio is up to 2 ppbv reflecting the
conversion of HC1 and C1ONO2 above. These results show
that by the end of December 1989, when large-scMe tem-
peratures inside the polar vortex had been cold enough for
PSCs, almost complete conversion of HC1 to active forms
could have occurred. Rood et al. [1992] discussed the possi-
bility that episodic dynamical events (which produce ozone
\"miniholes\") can cause a significant amount of processing
in December before the large-scale temperature falls below
the PSC threshold. Figure 1 shows that episodic process-
ing may have been important in early December 1989 (in
as much as the ECMWF analyses capture the episodic cold
temperatures) but by the end of December temperatures
were continuously well below the PSC formation threshold.
The OC10 molecule provides an important indication of
perturbed halogen chemistry in the polar stratosphere. It is
believed to be produced exclusively by the reaction
BrO + C10 - Br + OC10
During the day OC10 is rapidly photolyzed but at night
its concentration can build up. Perner et al. [1991] mea-
sured elevated levels of OC10 from January 5 to February 2,
1990, above Sondre Stromfjord (67øN, 51øW). Their results
confirm the OC10 observations of Solomon et al. [1988] in
1988 and Schiller ct al. [1990] in 1989. Pernor ct al. [1991]
measured a vertical twilight column of OC10 of between 0.8
and 1.6x10 3 molecules cm -2 (at 91 ø solar zenith angle) dur-
ing January and early February. Schiller ctal. [1990] mea-
sured a twilight vertical column of 1.0x10 a molecules cm -'
(at 88 ø SZA) and a nighttime column of 10x10 3 molecules
cm -9'. Figure 8a plots the column OC10 from the model
run B at 1200 UT on January 4. The OC10 column is large
inside the dark vortex with a maximum column of 8.0x10 3
molecules cm -'. Whilst a comparison with the twilight col-
umn measurements is difficult, the values in the model are
in reasonable accord with those obtained in the presence of
perturbed chemistry. The mixing ratio of OC10 at 500 K
on this day is 60-70pptv in the vortex as shown in Figure
8b.
Adrian ctal. [1991] made infrared measurements of the
column abundance of ItC1 from Esrange (68øN, 21øE), Swe-
den in January and February 1990. During January and
up to February 4 the measured vertical column was around
2.5x10 xs molecules cm -2. This corresponded to an es-
timated stratospheric column of 2x10 s molecules cm -:.
However, the next observations, on February 8 and 9, mea-
sured much larger ItC1 columns of 4.8x10 xs molecules cm -9'
o ,, ',, ,,..... , .:
a / ..... \".
: .;':.,, [ ,,---- ',. ,._. -, ,
EXP. B 90 1 4 12H
Fig. 8a. Column OOlO (x1013 molecules cm --2 ) from run B at
1200 UT on January 4.
CTM OCIO (ppfv) 500 K
EXP. B 90 1 4 12H
Fig. 8b. OC10 mixing ratio (pptv) at 500 K from run B at 1200
UT on January 4.
(equivalent o a stratospheric olumn of 4x10  molecules
cm-2). The increase from the 4,th to the 8th of Febru-
ary corresponded to warmer air moving over Esrange. As
Adrian et al. [1991] also measured the HF column, which
effectively remained constant through this period, they con-
cluded that before the 4th of February a significant frac-
tion of the stratospheric HC1 was converted to other ClOy

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY 7209
species or condensed in PSC particles. Figure 9 shows the
column ItC1 (above 300 hPa) on February 6 from run D.
The quantity plotted is the total H C1, taking into account
both gas phase and condensed HC1. With the PSC scheme
employed in the model, the amount of HC1 condensed in par-
ticles in regions with both type I and type II clouds would
be equivalent to a column of around 10 a4 molecules cm -2.
A local minimum of around 2.0 x10 as molecules cm -2 is sit-
uated to the east of Finland, coincident with the occurrence
of PSCs. This is surrounded by larger column abundances
of 3.5- 4.0x10 a molecules cm - in air which has not been
processed by PSCs. The large ]IC1 column in the model of
around 5.5x10 xs molecules cm -2 coincides with the warmest
lower stratospheric temperatures. This is a consequence of
the model initialization using 0 as the vertical coordinate;
where the atmosphere is warm the 2D model profile is de-
scended further. An inspection of the ECMWF analyses for
this period [Riishojsaard ½t al., 1992] shows that between
the 4th and the 10th of February the polar vortex shifted
towards Asia and rotated clockwise. The comparison of the
model results with the measurements are strongly influenced
by the model initial conditions and affected by the dynami-
cal evolution of the GCM forecast. However, the model sup-
ports the possibility that the movement of air over Kiruna
and the replacement of PSC-processed air by unprocessed
air could explain the increase in HC1 observed by Adrian ½t
al. [1991] in early February. A latitude altitude cross section
of C10 at 50øE is shown in Figure 10. The region of elevated
C10 in the polar lower stratosphere is clearly evident, with
mixing ratios of up to 1.4 ppbv at 40-50 hPa
Thus the presence of polar stratospheric louds severely
perturbs the partitioning of inorganic chlorine species from
that expected on the basis of ga phase chemistry. After
PSG processing, the time scale for the recovery of this gas
phase \"equilibrium\" isof the order of months [McKenna ½t
CTM Column HCI (10E 15 mole½.½m-2)
EXP. D 90 2 6 12H
Fig. 9. HC1 column (x10 x molecules cm -2) above 300 hPa from
run D at 1200 UT on February 5.
al., 1990]. First, as NO. is released by the photolysis of
HNOa it can react with C10 to form chlorine nitrate on a
time scale of a few days. Second, the recovery of HC1 by
reactions such as
C1 q- CH4 - HC1 q- CHa
takes place on a much longer time scale. The rate of this
recovery will therefore depend on the amount of insolation
and the extent of denitrification of an Mr parcel [Prather and
JafJfe, 1990]. In the 3D experiments described here, mixing
can also lead to the deactivation of C1Ox by the mixing of
mid-latitude and polar air.
Bromine species. Figure 11a plots the mixing ratio of
BrONO2 at 50 hPa on February 6. The 'denoxification\"
(conversion of NOx species to HNO) of the air in the pres-
ence of PSCs has lead to a large reduction of BrONO2 over
northern Europe and northern Russia. Figure 11b shows
that BrO is now the dominant bromine species in the sunlit
atmosphere over Scandinavia with a peak mixing ratio of 7
pptv. In the region of perturbed air which is in darkness,
BrC1 is the main nighttime reservoir (not shown). A latitude
height cross section at 50øE of BrO is shown in Figure 12
for 1200 UT on February 6th. The region of PSC-perturbed
chemistry in the Arctic lower stratosphere is clearly marked,
with elevated BrO levels between 50-70øN and 100-20 hPa.
At 80 hPa (approximately 0=400 K) there is a very strong
vertical gradient of BrO which increases to over 7 pptv be-
tween 60-20 hPa (approximately 0=440-580 K). During the
AASE, Toohey ½t al. [1990] measured BrO mixing ratios
of 44-2 pptv at 0=400 K rising to 84-2 pptv at 0=470 K,
which is in agreement with the model. The integrated col-
umn of BrO in the model tends to underestimate the avail-
able observations, however. During the AASE Wahucr ct
al., [1990] generally measured vertical BrO column densities
from 2-5x10 3 molecules cm -2 and on one occasion mea-
sured a column of 134-4x10 xa molecules cm -e. From Sondre
Stromi]ord Perncr ½t al. [1991] measured vertical column
BrO amounts of 4-9x10 3 molecules cm -2. A plot of col-
umn BrO from run D is shown in Figure 11c. The rnaximum
column is 2.2x10 x3 molecules cm -e which with a mixing ra-
tio of BrO of 8 pptv at 30 hPa decreasing to 6 pptv at 70
hPa and 4 pptv at 100 hPa. The mixing ratio of inorganic
bromine in the upper stratosphere in the model is 12 pptv,
which is probably an underestimate.
Ozone Destruction. The model experiments have been
analyzed to assess the the magnitude of the chemical ozone
destruction that may have occurred during winter 1989-
1990. Comparison of the model ozone from runs A and
B gives an indication for the potential for ozone destruction
in the vortex during the midwinter period. Figure 13 plots
the difference in O3 between run B and run A on the 500 K
isentropic surface on the January 4 1990. There is very lit-
tle ozone loss in run B; at the edge of the polar vortex peak
destruction of 0.015 ppmv (an average of 1.5 ppbv/day) has
occurred. The lack of sunlight at this time of year limits the
destruction to a ring near the polar night terminator.
Comparison of runs C and D gives an estimate of the
ozone destruction expected as a result of the PSC event
of early February 1990. Throughout the period of these
simulations the coldest stratospheric temperatures, where
PSCs are possible, remained over northern Scandinavia and
the northern Soviet Union. Thus circulation within the po-
lar vortex would transport air through polar stratospheric

7210 CHIPPERYIELD El' AL.' THREE-DIMENSIONAL MODELINO STUDY
2-
7-
10-
4O
7O
1 oo
2OO
400 -
700 i 1000
-90
CTM 010 (ppbv)
0..2 
-80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90
Latitude (degrees)
Fig. 10. Latitude cross section (at 50ø]) of the mixing ratio of C10 (ppbv) from run D at 1200 UT on February
6. Contour interval is 0.2 ppbv.
CTM BrON02 (pptv) 50 hPa CTM BrO (pptv) 50 hPa
':\".L,-' ', ,::-.
..... !:--..':--:? ::¾:::.:: ,:
EXP. D 90 2 6 12H
Fig. 11a. Mixing ratio of BrONO2 (pptv) at 50 hPa at 1200 UT
on February 6 from run D. Contour interval is I pptv.
EXP. D 90 2 6 12H
Fig. 11b. As Figure 11a but for BrO.
clouds implying that chemical O3 destruction should be ob-
served in and downwind of this region. Figure 14a plots
the difference in O3 mixing ratio between run C and run
D at 50 hPa on February 14th. Within the polar vortex
ozone destruction of up to 150 ppbv has occurred over 10
days; an average of 15 ppbv/day or around 0.75%/day. The
corresponding loss in the column amount is around 7 D U
(Figure 14b). Despite the high levels of active chlorine in
the model, the O3 loss is still quite limited due to the low
amount of sunlight in the northern polar region in early
February. During winter 1989-1990 early February would
have provided the best opportunity for the coincidence of

CHIPPERFIELD El' AL.: THREE-DIMENSIONAL MODELING STUDY 77.,11
CTM Column BrO (10E13 molec.cm-2)
EXP. D 90 2 6 6H
ß / .
:,,/ ........ :,
,.,
tentiM for ozone loss if PSCs persist until early March. Fig-
ure 15a shows that after 10 days ozone loss of 250 ppbv has
occurred in the vortex at 50 hPa. The largest loss in the
column (Figure 15b) is 14 DU near 70øN, 120øE. The local
loss rate at 50 hPa of 25 ppbv/day (around 1.25%/day) in
early March is similar to the rate of ozone loss observed at
18 km in early September in the Antarctic [Anderson et al.,
1991]. In the future significant ozone depletion could occur
in the Arctic in winters with a late final warming.
An average ozone loss of 15 ppbv/day at 50 hPa in run
D is similar to, but slightly less than that inferred by Koike
et al. [1991]. It is also similar to the gradual trend seen by
Ho/mann and Deshler [1991] although it is much less than
the rate of episodic loss inferred by them. The magnitude
of ozone loss calculated with the model in early February
(experiment D) is much less than that inferred by Le/gvre et
al. [1991]. The large discrepancy between their model ozone
field and the TOMS data is inconsistent with the chemical
destruction mechanisms analyzed here. Model run D used
the same initial conditions for ozone and the meteorological
variables as the high resolution (T79) experiments of Le/dvre
et al [1991] and Riishojgaard et al. [1992]. Overall, when
compared to TOMS data (not shown), the evolution of the
column ozone in run D is poorer than that by Riishojgaard
et al. [1992], as would be expected with the lower resolution
Fig. 11c. Column BrO (x1013 molecules cm -2) at 0600 UT on
February 6 from run D.
PSC-processed air and sunlight. The relatively small mag-
nitude of the ozone loss predicted in the model, compared
to the variability of ozone induced by dynamics, highhghts
the problems in identifying the signature of chemical ozone
loss in the Arctic.
Experiments E and F, which used the same dynamical
forcing as experiments C and D, indicate the increased po-
70
1 O0
200
7-
lO--
20-
40-
400
CTM
I
9
of T21. However, the gross features in the model ozone
field are similar as the experiments were performed with
essentially the same dynamical model. Le/gvre et al (1991)
argued that the initial ozone field synthesized from TOMS
data provides a good estimate of the profile. If the evolution
of the model dynamics is realistic then this would suggest
that the large discrepancy in the ozone column between their
T79 experiments and the TOMS data over central Siberia
is due either to an underestimate by the TOMS instrument
or a rapid, as yet unknown, chemical ozone sink. However,
BrO (pp'tv)
700
1000 ' I ' I ' I ' I ' t ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I '
-90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80
Latitude (degrees)
Fig. 12. Latitude cross section (at 50øE) of the mixing ratio of BrO (pptv) from run D at 1200 UT on February
6. Contour interval is I pptv.

7212 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELINO STUDY
/ø . )::7 ß ß -
,.,' . _ ............. 7:::.,.)..- - - ,, ... '-.- _ \" ,. -:,-- dZ _,, I ,, ..... '\" ...... ~'\" .....
EXP. B- A 90 1 4 12H
Fig. 13. The difference in 03 mixing ratio (ppbv) between run B
and run A on January 4 1990.
CTM DIFFERENCE COLUMN 0.3 (DU)
\"% i /' \"':\"\"'2, x N ß
,.\" \",. ...'! . -/,-- ;::,:\"\"'\"%-:\";::'  ,.....I
[:\" ...... \"\"' ...... ':'/\";':\"\"' ..... \"::' '\"' : :\"'1 ...... \"> .... \", .? ! :' ,'\" \"'>>\" \"'\"'- ::5' ....... ,, \".',, '. ..  .. ':,\" .: .... '[\":\"-'-:.,'Z: ',......,', ' , .. ' '.. ./. . \"\"-:. ... ... ','.}.'-':':' .  .'.::\" .- ? 'i-' '\"'
i '\"... / .:¾-'\" \" \",
ß ' .. 2::'-. :,/' '.,, .......
ß ...... .' .,\"',-,.,,'-.'\" / 
.................. -,' >.. ;- {..-..,..- .'::','z ..................................................
:'\":':.,' ' ! '.,'!i53 :Z'\" %-L.' . ..
','.'..'¾'* .<!:'.,:. \"¾:[ 'i'>.!':'2\" ,. >' ½:., \"% ,
' .... i
..... :.' .:i\"'?i '--'.\"'\"\"':\"\"--- .... :i_ . ' ...... .:' _?-..
. , ,, .,, ,
; ..... j: ./.'-.: ....... .. ,:,:. x,....,.'\", ß : ,  ...... ..:.,..,.. .... ,-...._.. '.,, /'
ß . ,/ ,. -.,..-., .....  \ ,
,-\"  ,- .: -..-F \",,-., ', ....... ,/
b '/' \"-':' ' ..... \"'\"' ' ......... :;\"'<:?' ....
' .,:\"'l L. .....  .,-' '-,, \"''
/ ,.'\": ..... ';', ..... I .... - ........ ' \" .... \"- (
EXP. D - C 90 2 14 12H
Fig. 14t. ?he dfference in colum O (DU) between ru D ad
run C on February 14 1990.
CTM DIFFERENCE 03 (ppmv) 50 hPo
EXP. D - C 90 2 14 12H
Fig. 14. \"I'he difference in 03 mixing rt, io (ppm¾) t I.-50 ]-tP
between run D and run  on February 14 1990.
CTM DIFFERENCE 03 (ppmv) 50 hPo
' \"\ I / --,
: I'
':  '\"'-,.,,.
/ \ ]1 ,/' ,..-'\"\";% 2, '\".. - ' ß \", ß: -/..,.-....'c'\":)/'\"\"'\":'>:? x ...
....... ß ',,. :,; , i.i '- '\ .? I ' \"'\"f' '\"\ .... :;'.'','-- ', .f;i ,. :-- ..... ::..7':':.;'Z ...... '\"..,. ¾;,,'\" 2.,  / .,::,,.. ,,.,:..,..:, . ',, ':-.: {\".,1 ,. \
:d':':': \"' ' % :' '\"..
' ...... ;'.Z.::': ...... ; ..... i\"\".g': ?:-':.:'-'\" ''  ...........................
ii:] ::::&.:..
' '\":'\":::'::i':: ...... : ,' .... ,,.,,....,, '\"'\" ... .....  \"' V '.. .'.
....... ß ':' .,.:.::...,...:' _ .:::: _..:: : ::, : .. \"'
;/'
EXP. F - E 90 Z 14 OH
. 1. he difference i Os m/xin rtio (ppmv) t 0 hP
between run  and run B on Mach 14 1990.
uncertainties in the reconstructed ozone profile could be the
source of the discrepancy.
The chemical cycles responsible for this ozone destruction
have been analyzed. McKenna et al. [1990] used results
from an isentropic trajectory model at 470 K and concluded
that the cycle involving ClO + ClO was by far the most
significant loss cycle in early February. In the study of Mc-
Cormell et al. [1991] loss due to the cycle involving C10 +
O dominated at 20 km. Murphy[1991] used ER-2 data from
the AASE campaign to calculate the expected loss due to
the cycles involving C10 q- C10 and C10 q- BrO over the
aircraft flight range. ¾e have used our 3D model results
to investigate how the efficiencies of these catalytic cycles
vary with altitude, latitude and time. The model conti-

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELINO STUDY 7213
CTM DIFFERENCE COLUMN 0;5 (DU)
EXP. F - E 90 2 14 OH
Fig. 15b. The difference in colman 03 (DU) between run F and
run E on March 14 1990.
nuity equation for O was combined with the steady state
expressions used to partition the model famihes to identify
the rate determining steps in the important catalytic cycles
[Johnston. and Podolskc, 1978]. In model runs D and F the
following cycles dominate the ozone loss in the polar lower
stratosphere:
Cycle I
Cycle II
CiO + ClO + M  C1202 + M
01202 -Jr- hu  0102 + C1
C102 -Jr- M  C1 + 02 + M
2(O1+O3 --, 010+O2)
net: 2Ca --* 302
C10+BrO --+ Cl+Br+02
Cl+O3 --* CIO+O2
Br+O3 - BrO+02
net: 203 - 302
Cycle II can also be initialized by the formation and subse-
quent photolysis of BrC1.
Cycle III CiO+O - C1+O2
Cl+O3 - C10-+-O2
net ß 0 + 03  202
Cycle IV CiO + HO2 -
HOCi + hv -
Ci + O3 -
OH +Ca -
net: 203 ---
HOC1 + 02
OH +C1
C10 + 02
HO2 + 02
302
The other catalytic cycles make a minor contribution to the
ozone loss. Bekki et al. [1991] noted that in the presence of
heterogeneous chemistry on sulphate aerosols ozone loss in
the lower stratosphere is controlled mostly by HO chem-
istry. The \"chronic\" effects of heterogeneous processing on
sulphate aerosols can be contrasted with the \"acute\" effect
of processing by PSCs where ozone loss is dominated by
the C10 and BrO radicals. Figure 16 plots the ozone loss
due to cycle I in early February and early March. In gen-
eral, this cycle is the dominant loss cycle in both runs D
and F throughout the winter. In early February (Figure
16a) the maximum instantaneous loss rate (at about 1500
local time) is 50 ppbv/day centered at 60øN and between 35
and 50 hPa. By early March the peak loss rate is over 75
ppbv/day. The efficiency of cycle I increases at low altitudes
as it is initiated by the three-body formation of C1202. At
altitudes below 60 hPa the efficiency of the catalytic cycles
is limited by the availability of active chlorine (see Figure
10) which in turn is limited by the total inorganic chlorine.
It is ozone destruction at the lower altitudes which will have
the largest impact on the integrated column amount. Figure
17 shows the ozone loss due to cycle II. In early February
the maximum loss is 25 ppbv/day at around 30-50 hPa at
55øN. Thus the peak destruction due to cycle II is about
50% of that due to cycle I. The importance of cycle II could
6-
8-
_
10-
2O-
d[OS]/df due to 010 + ClO ppbv/doy
Latitudo
d[OS]/df due fo ClO + CIO ppbv/doy
2øø15 ' G' L,' ;,' 15' ;o ' ,' ' io '  ' .o
Latitude (degrees)
Fig. 16. Instantaneous rate of loss of O3 (ppbv/day) at 50øE at
1200 UT (about 1500 local time) due to catalytic cycle I for a)
February 6 from run D and b) March 6 from run F.

7214 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELINC] STUDY
d[OS]/df due to CIO + BrO ppbv/day
45 50 55 60 65 70 75 80
Lnfitude (degrees)
i
d[OS]/df due to CIO + BrO ppbv/doy
lO
'- -20 -'10
Latitude (degrees)
Fig. 17. As Figure 16 but for cycle II.
be underestimated, however, as the model runs contained
a peak stratospheric mixing ratio of inorganic bromine of
12 pptv. By early March the loss due to cycle II has only
increased to just over 30 ppbv/day. Now the contribution
of cycle II is less compared to cycle I (about 40%). Al-
though the rate of ozone loss due to cycle I is much greater
than that due to cycle II at the peak of the C10 mixing
ratio (Figure 10) near the periphery of the region of ele-
vated C10 and BrO (Figure 12) loss due to the two cycles
is similar. The location of the-10 ppbv/day contour is es-
sentially the same in Figures 16a and 17a. The peak in BrO
mixing ratio is much broader than the peak in C10 mixing
ratio. Ozone loss due to cycle III is plotted in Figure 18.
In early February the peak ozone loss rate is 55 ppbv/day
situated around 28 hPa and 53-57øN. At 50 hPa the loss due
to cycle III is only 10-15 ppbv/day. Thus when expressed
as the mixing ratio loss, cycle III can destroy ozone at a
rate comparable to, or even greater than, cycle I. However,
the destruction due to cycle II! is a maximum above 30 hPa
due to the rapid increase of atomic oxygen with altitude.
Ozone loss at around 20-30 hPa will have less of an effect
on the column and can also be compensated for, to some
extent, by in situ production of odd oxygen (see below). By
early March the maximum ozone loss due to cycle III is still
around 50 ppbv/day over a similar altitude range but now,
due to the increased sunlight, loss occurs further towards
d[O.3]/dt due to ClO + 0 ppbv/day
/ I I I t I
10
13_ -
80-
45 55 60 70 75
Latitude (degrees)
I i
d[O$]/dt due to CIO + 0 ppbv/doy
i i , i , i , i
45 50 55 0 85 70
Latitude (degrees)
Fig. 18. As Figure 16 but for cycle III (N.B. contours stop at
-80).
the pole. Results for cycle IV are shown in Figure 19. In
early February ozone loss due to this cycle is negligible. In
early March the maximum loss of 25 ppbv/day occurs at
around 30 hPa. The large increase in the efficiency of cycle
IV reflects the increase in H O2. In situ ozone production
due to the photolysis of Os is shown in Figure 20. At alti-
tudes above 30 hPa where cycle III is efficient at destroying
ozone the photolysis of Os can compensate for the loss. At
30 hPa and 60øN in early March the rate of in situ ozone
production is 30 ppbv/day. Climatology [e.g., Barnett and
Corney, 1985] shows that temperatures are warmer in the
Arctic winter stratosphere compared to the Antarctic. This
limits the most likely occurrence of PSCs in the northern
hemisphere to the altitude range 18-21 km (see, e.g. World
Meteorological Organisation/UNEP, [1991]). In the south-
ern hemisphere PSCs can commonly occur down to altitudes
as low as 14 km and below. Therefore in the Arctic, chlo-
rine activation by PSCs may be largely limited to higher
altitudes than in the Antarctic where the efficiency of cycle
I (the main cause of ozone loss in the Antarctic ozone hole)
is reduced.
The evolution of the potential for ozone destruction by
cycles I-III during the Arctic winter is illustrated by Figure
21. These data were calculated with a simple box model and
shows the integrated amount of ozone destruction per day

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELINC; STUDY 721 $
d[O3]/dt due to ClO + H02
lO
),,o-
60-
_
80-
_
100-
45
ppbv/day
I I
Latitude (degrees)
d[OS]/dt due to J[02] ppbv/day
80-
.
1oo-
200 I ' I ' 1 ' I ' I ' ;5 ' I ' I '- 45 50 5 60 8B 70 80 85 90
Latitude (degrees)
d[O3]/df due to ClO + H02
10 
ppbv/day
I
Latitude (degrees)
Fig. 19. As Fibre 16 buC for cycle IV.
for these three cycles from January to March for an air par-
cel at 70øN and 50 hPa. The calculation assumes that the
air parcel contains 1.5 ppbv of active chlorine and 10 pptv
of BrO. During January the loss due to cycle I and cycle
II is similar and increases to just under 2 ppbv/day at the
end of the month. By the end of February, cycle I destroys
11-12 ppbv/day and by the end of March over 20 ppbv/day.
Destruction by cycle II increases to 13 ppbv/day by the end
of March. Also shown in the figure is the hours of sunlight
per day at this point in the atmosphere. The efficiency of
cycle I increases more rapidly over this period than cycles II
and III. As the concentration of daytime BrO is effectively
constant in the box the increase in ozone destruction due
to cycle II reflects the shift of the C10 equilibrium from
C1202 to C10 as the intensity of sunlight increases and also
the increasing amount of daylight. Similarly, as the con-
centration of O atoms at this altitude is mainly controlled
by photolysis of O3 in the Chappuis band, where the atmo-
sphere is optically thin, the mean daytime O concentration
will not change by much over this period. The increase in
O3 loss due to cycle III therefore also reflects the increase
in daytime C10 and the increasing amount of daylight. At
altitudes above 30 hPa, where cycle III is most important,
the balance between C10 and its dimer will be shifted more
towards C10 as the intensity of sunlight is larger and the
three-body formation of C1202 is slower. Therefore the loss
d[O3]/dt due fo J[02] ppbv/day
60-
80-
.
lOO-
200 ' i , I , i ......
Latitude (degrees)
Fig. 20. As Figure 16 but for production of odd oxygen by the
photolysis of 02 (N.B. irregular spacing of contours).
of ozone due to cycle III will depend almost entirely on the
length of the day, as shown in Figure 18. The increase in
ozone loss due to cycle I is faster due to the quadratic ef-
fect of the increase of daytime C10 on the rate of this cy-
cle. Figure 21 illustrates the large increase in potential for
ozone destruction if the polar vortex remains cold and sta-
ble beyond early February, the case in 1990. The increasing
ozone destruction with time shown in the figure could also
be achieved by the air parcel traveling equatorwards where
more sunlight is available.
Figures 16-21 illustrate the different efficiencies for cat-
alytic cycles in the destruction of ozone in the northern and
southern polar regions. In the formation of the Antarctic
ozone hole the O3 loss occurs in September and is due mainly
to the C10 dimer cycle with a minor role (25%) for bromine
[Anderson ½t al., 1991]. This season corresponds to March
in the northern hemisphere where the dominance of cycle I
is shown in Figures 16 and 21. However, by March tempera-
tures are usually too warm for Arctic PSCs. In January and
early February when Arctic PSCs are most common, the cy-
cles involving BrO are relatively more important. However,
for a strong Arctic ozone depletion, requiring PSCs to per-
sist until late March, the C10 dimer will again be the most
important cycle.
5. CONCLUSIONS
A three-dimensional radiative-dynamical-chemical model

7216 CHIPPERFIELD ET AL.' THREE-DIMENSIONAL MODELINO STUDY
Ozone production/destruction
22 ,,... -
20 ..-' -
Cycle T ,/'
/..
!8 // -
./
,16 ,/- - -,-,
\" 14- ..-.-'\" -,,
.(:3 //\"
,_.
c:z. 12 / .....
o3 lO ß ..--  - 0
,.-.., .....,,...:,:,::/ ,., ycle I I .., o
\"0 8 .,--' ,,// , ß '' ..   *\" -
ß
\" /\" .,\" - .... Cyc'le Z ZZ - 6 daylight
4 . ½\"-' ' ...... J[02]
,.r..--l-  ...... ..,-, ,, .,-- , ...,....,.....,....,..,.,,--..-..-.-
[   'r*- \"*'r o \"\"1 ;' ':::f:\"-:'=:-r .....  ..... 1 ......... 1- ...... t-:=-:\"-'fi =\"-'- ..... r .... T-' -r-- - .... -r- --T-- -r ....  ....
0 5 10 15 20 25 80 85 40 45 50 55 60 65 70 75 80 85 90
Day of the year
:Fi$. 21. R, al;e of loss of 03 (ppbv/clay) clue to cycles I II az'td III for az't a3r paxce] a.t, 50 :h:Pa, 70øN &ssumln$ a.
rrdxin$ ra.to of C]O 1.5 ppbv BrO 10 ppt¾ O3 2.5 pprnv a.t T=200 K. Also s:hown is Lhe rate of odd oxyse
procluclfio clue to l;:he p:hotolysis of 02 axtcl l;:he umber oœ :hours of sunsl'Lie per cl&y.
has been used to study the distribution of trace gases in
the Arctic lower stratosphere during winter 1989-1990. The
model was initialized using meteorological analyses of the
ECMWF and using constituent data from a 2D model trans-
formed into the coordlnates of PV and O. In some exper-
iments TOMS data were used to initialize the model O3
field. The model contains a comprehensive description of
gas phase chemical reactions as well as a treatment of type
I and type II polar stratospheric clouds. In the CTM the
chemical species are transported using the second-order mo-
ments scheme of Prather[1986]. The use of this non diffusive
scheme has resulted in a great improvement over the previ-
ous spectral scheme. This is iSarticUlarly important in the
comparatively low resolutions which the cost of 3D chemical
transport models often necessitates.
A series of 10-day experiments were performed through-
out the 1989-!990 winier. Type I PSCs would have begun
to form around mid December. By the end of December the
model predicts significant processing of air with a large re-
duction in HC1 and an associated increase in active chlorine.
The period of January 1990 was very cold with tempera-
tures consistently below the threshold for the formation of
type I PSCs in the polar vortex. In late January and early
February type II PSCs were able to form. The chemical
ozone loss produced in the model varied strongly with time
through the amount of sunlight available. In early January
the ozone loss (around 1.5 ppbv/day at 50 hPa) was small
and confined to the edge of the vortex, near the terminator.
In early February the rate of ozone loss in PSC-processed air
was 15 ppbv/day (around 0.75%/day) at 50 hPa correspond-
ing to a loss of just under 1DU/day from the column. In
mid February 1990 temperatures became too.. warm for fur-
ther PSC activity. If PSCs had persisted until early March
ozone loss of around 25 ppbv/day at 50 hPa could have been
sustained,. In the future significant ozone depletion could oc-
cur for winters with late final warmings.
The cycles responsible forthe destruction (f O3 have been
analyzed using the 3D model results as a function of lati-
tude, altitude and time. In general, the cycle initiated by
the reaction C10 + C10 is the dominant loss mechanism be-
low 30 hPa. However, early in January the cycles involving
C10 + BrO become relatively more important. This would
be especially true for BrO levels greater than 10 pptv. Also,
where C10 levels are low, e.g., at the edge of the chemically
perturbed region, the cycles involving bromine will play a
relatively more important role. Later in the season (corre-
sponding to the conditions in which the Antarctic ozone hole
forms) the increasing intensity of sunlight, and the resultant
shift of the equilibrium of C1202 to C10 greatly increases the
efficiency of the C10 + C10 cycle. At altitudes above 30 hPa
the cycle involving the reaction C10 + O is most efficient at
destroying ozone. At these altitudes some in situ compen-
sation for the O3 loss can occur through the photolysis of
02.
A number of trace species in the model have been com-
pared with the available observations. In particular, results
from the model are in accord with the ground based obser-
vations of I-INO3, NO2, HC1 and OC10. The model (with 12
pptv of inorganic bromine in the upper stratosphere) under-
estimates the measurements of the BrO column. The avail-
able measurements are, however, too sparse to effectively
constrain the 3D model.

CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY 7217
Although the results from the 3D model are encouraging
and in agreement with observations a number of uncertain-
ties remain. Short simulations of a 3D model are heavily
influenced by the initial chemical conditions. Ideally, global
constituent fields should be used to initialize and constrain
the model. The experiments presented here were performed
at the low horizontal resolution of T21 (about 5 ø x 5 ø) due to
the computational expense of the detailed chemistry scheme.
The winds and temperature used for the CTM were taken
from a GCM simulation which also used the resolution of
T21. In this low resolution the meteorological evolution of
the model will diverge from reality quite quickly. A better
forcing for the CTM would be obtained by using winds from
a higher resolution GCM experiments or from assimilated
data.
Acknowledgments. M.P.C. thanks NATO for funding a re-
search fellowship which is administered by the U.K. Science and
Engineering Research Council. This work was sponsored by
the Programme Atmosphere Moyenne of the Centre National
d'Etudes Spatiales, the Centre National de la Recherche Scien-
tifique and the Ministre de l'Environnement (under grant 91-
100), and by the Commission of the European Communities (un-
der STEP project 016).
REFERENCES
Abbatt, J.P. D., K. D. Beyer, A. F. Fucaloro, J. R. McMahon,
P. J. Wooldridge, 1. Zhang, and M. J. Molina, Interaction of
HC1 vapor with water-ice: Implications for the Stratosphere,
J. Geophys. Res., 97, 15,819-15826, 1992.
Adrian, G. P., T. Blumenstock, H. Fischer, L. Gerhardt, T.
Guide, H. Oelhaf, P. Thomas, and O. Trieschmann, Column
mounts of trace gases derived from ground-based measure-
ments with MIPAS during CHEOPS III, Geophys. Res. Left.,
18, 783-786, 1991.
Anderson, J. G., D. W. Toohey and W. H. Brune, Free radicals
within the Antarctic vortex: The role of CFC's in the Antarctic
ozone loss, Science, 251, 39-46, 1991.
Austin, J., R. R. Garcia, J. M. Russell Ill, S. Solomon, and A.
F. Tuck, On the atmospheric photochemistry of nitric acid, J.
Geophys. Res., 91, 5477-5485, 1986.
Barnett, J. J., and M. Corney, Middle atmosphere reference
model derived from satellite data, Middle Atmosphere Pro-
gram, Handbook for MAP, vol. 16, pp. 47-85, Scientific Com-
mittee on Solar-Terrestrial Physics (SCOSTEP), Copies avail-
able from SCOSTEP Secretariat, University of Illinois, 1985.
Bekki, S., R. Toumi, J. A. Pyle, and A. E. Jones, Future aircraft
and global ozone, Nature, 354, 193-194, 1991.
Cariolle, D., and D. Brard, The distribution of ozone and other
active species : results of a two-dimensional model, in Pro-
cedings of the Quadrennial Ozone Symposium, edited by R.
D. Bojkov and P. Fabian, Halkidiki, pp77-81, D. Reidel, Dor-
drecht, 1984.
Carlolie, D., A. Lasserre-Bigorry, J. F. Royer, and J. F. Geleyn, A
general circulation model simulation of the springtime Antarc-
tic ozone decrease and its impact on mid-latitudes, J. Geophys.
Res., 95, 1883-1898, 1990.
DeMote, W.B., et al, Chemical kinetics and photochemical
data for use in stratospheric modeling, Evaluation no. 9,
NASA/JPL Publ. 90-1, Jet Propul. Lab., Pasadena, Calif.,
1990.
Douglass, A. R., R. B. Rood, R. S. Stolarski, M. R. Schoeberl,
M. H. Proffitt, J. J. Margitan, M. Loewenstein, J. 1. Podolske,
and S. E. Strahan, Global three-dimensional constituent fields
derived from profile data, Geophys. Res. Left., 17, 525-528,
1990.
Douglass, A. R., R. B. Rood, J. A. Kaye, R. S. Stolarski, D. J.
Allen and E. M. Larson, The influence of polar stratospheric
heterogeneous processes on reactive chlorine at middle lati-
tudes: Three-dimensional model implications, Geophys. Res.
Left., 18, 25-28, 1991.
Granier, C., and G. Brasseur, Ozone and other trace gases in
the Arctic and Antarctic regions: Three-dimensional model
simulations, J. Geophys. Res., 96, 2995-301!, 1991.
Hanson, D., and K. Mauersberger, Laboratory studies of the ni-
tric acid trihydrate: Implications for the south polar strato-
sphere, Geophys. Res. Lett., 15, 855-858, !988a.
Hanson, D., and K. Mauersberger, Solubility and equilibrium va-
por pressures of HC1 dissolved in polar stratospheric cloud ma-
terial: Ice and the trihydrate of nitric acid, Geophys. Res. Lett.,
15, 1507-1510, 1988b.
Harwood R. S. and J. A. Pyle, A two-dimensional mean circula-
tion model for the atmosphere below 80kin, Q. J. R. Meteol.
Soc., 101, 723-748, 1975.
Holmann, D. J., and T. Deshler, Evidence from balloon measure-
ments for chemical ozone depletion of stratospheric ozone in
the Arctic winter of 1989-90, Nature, 349, 300-305, 1991.
Johnston, H. S., and J. Podolske, Interpretation of stratospheric
photochemistry, Rev. Geophys., 16, 491-519, 1978.
Iaye, J. A., A. R. Douglass, R. B. Rood, R. S. Stolarski, P. A.
Newman, D. J. Allen, and E. M. Larson, Spatial and temporal
variability of the extent of chentically processed stratospheric
air, Geophys. Res. Left., 18, 29-32, 1991.
Ioike, M., Y. Iondo, M. Hayashi, Y. Iwasaka, P. A. Newman,
M. Helten, and P. Aimedieu, Depletion of Arctic ozone in the
winter 1990, Geophys. Res. Letg., 18, 791-794, 1991.
Lary, D. J., and J. A. Pyle, Diffuse radiation, twilight and pho-
tochemistry, J. Atmos. Chem., 13, 373-392, 1991.
Lef[vre, F., L. P. Riishojgaard, D. Carlolie, and P. Simon, Mod-
elling the February 1990 PSC type II event and its potential
impact on the Northern hemisphere ozone content, J. Geophys.
Res., 96, 22,509-22,534, 1991.
McConnell, J. C., W. F. J. Evans and E. M. J. Templeton, Model
simulation of chemical depletion of Arctic ozone during the
winter of 1989, J. Geophys. Res., 96, 10,923-10,930, 1991.
McI<enna, D. S., 1. L. Jones, L. R. Poole, S. Solomon, D. W. Fa-
hey, I<. I. Ielly, M. H. Proffitt, W. H. Brune, M. Loewenstein,
and I. R. Chan, Calculation of ozone destruction dm'ing the
1988/89 Arctic winter, Geophys. Res. Left., 17, 553-556, 1990.
Muller, R., The performance of classical versus rnodern finite
one-dimensionaltest-bed, Mon. Weather Rev., 120, 1407-1415,
1992.
Murphy, D. M., Ozone loss rates calculated along ER-2 flight
tracks, J. Geophys. Res., 96, 5045-5053, 1991.
Murray, F. W., On the computation of saturation vapor pressure,
J. Appl. Meteorok, 6, 203-204, 1967.
Naujokat, B., I. Labitzke, R. Lenschow, K. Petzoldt, and R.-C.
Wohlfart, The stratospheric winter of 1989/90: Very cold with
a pronounced minor warming and a late final warming, Bell.
Berl. Wetterkarte, 75/90, S012/90, Freien Universitat Berlin,
1990.
Perner, D., T. Ilupfel, U. Parchatka, A. Roth, and T. Jor-
gensen, Ground based UV-vis spectroscopy: Diurnal OC10
profiles during January 1990 above Sondre Stromfjord, Green-
land, Geophys. Res. Left., 18, 787-790, 1991.
Pommereau, J.P., F. Goutail, and A. Sarkissan, Observations
de la stratosphere par spectrometrie UV-visible depuis le sol,
PAMOY CHEOPS IlI, final report, edited by J.P. Pom-
mereau, Verrieres le Buisson, 1991.
Pommereau, J.P., and U. Schmidt, CHEOPS III: An ozone re-
search campaign in the Arctic winter stratosphere 1989/90,
Geophys. Res. Left., 18, 759-762, 1991.
Prather, M. J., Numerical advection by conservation of second-
order moments, J. Geophys. Res., 91, 6671-6681, 1986.
Prather, M., and A. H. Jaffe, Global Impact of the Antarctic
ozone hole: chemical propagation, J. Geophys. Res., 95, 3473-
3492, 1990.
Ramaroson, R. A., Modelisation locale, a une et trois dimensions
des processus photochimiques de l'atmosphere moyenne, Ph.D.
thesis, Univ. Paris VI, 1989.
Ramaroson, R., D. Cariolle and M. Pirre, Simulation of Noxon
cliff in a three-dimensional model with diurnal variations dur-
ing an unperturbed winter, Polar Stratospheric Ozone, in Pro-
ceedings of the First European Workshop, 3-5 October 1990,
Schliersee, FRG, edited by J. A. Pyle and N. R. P. Harris,
pp267-270, Commission of the European Communities, 1990.
Ramaroson, R., M. Pirre, and D. Cariolle, A box model for on-line
computations of diurnal variations in a 1D model: Potential
for application to multidimensional cases, Ann. Geophys., 10,

7218 CHIPPERFIELD ET AL.: THREE-DIMENSIONAL MODELING STUDY
416-428, 1992.
Rattigan, O., E. Lutman, R. L. Jones, R. A. Cox, K. Clemitshaw,
and J. Williams, Temperature dependent absorption cross sec-
tions of gaseous nitric acid and methyl nitrate, J. Photochem.
Photobiol. A, 66, 313-326, 1992.
Riishojgaard, L. P., F. Lefevre, D. Cariolle, and P. Simon, A
GCM simulation of the northern hemisphere ozone field in
early February 1990 using satellite total ozone for model ini-
tialisation, Ann. Geophys., 10, 54-74, 1992.
Rodriguez, J. M. et al., Nitrogen and chlorine species in the spring
Antarctic stratosphere: Comparison of models with airborne
Antarctic ozone experiment observations, J. Geophys. Res., 94,
16,6883-16,703, 1989.
Rood, R. B., J. A. Kaye, A. R. Douglass, D. J. Allen, S. Steenrod
and E. M. Larson, Wintertime nitric acid chemistry: Impli-
cations from three-dimensional model calculations, J. Atrnos.
Sci., 4 7, 2696-2709, 1990.
Rood, R. B., J. E. Nielsen, R. S. Stolarski, A. R. Douglass, J. A.
Kaye, and D. J. Allen, Episodic total ozone minima and associ-
ated effects of heterogeneous chemistry and lower stratospheric
transport, J. Geophys. Res., 97, 7979-7996, 1992.
Salawitch, R. J., et al, Loss of ozone in the Arctic vortex for the
winter of 1989, Geophys. Res. Left., 17, 561-564, 1990.
Schiller, C., A. Wahner, U. Platt, H.-P. Dom, J. Callies, and D.
H. Ehalt, Near uv atmospheric absorption measurements of
column abundances during airborne Arctic stratospheric expe-
dition, January- February 1989, 2, OC10 observations, Geo-
phys. Res. Lett., 17, 501-504, 1990.
Schmidt, U., R. Bauer, A. Khedim, E. Klein, G. Kulessa, and
C. Schiller, Profile observations of long-lived trace gases in the
Antarctic vortex, Geophys. Res. Left., 18, 767-770, 1991.
Schoeberl, M. R., M. H. Proffitt, K. K. Kelly, L. R. Lair, P. A.
Newman, J. E. Rosenfield, M. Loewenstein, J. R. Podolske, S.
E. Strahan, and K. R. Chan, Stratospheric constituent trends
derived from ER-2 profile data, Geophys. Res. Left., 17, 469-
472, 1990.
Solomon, S., Progress towards a quantitative understanding of
Antarctic ozone depletion, Nature, 34 7, 347-354, 1990.
Solomon S., G. H. Mount, R. W. Sanders, R. O. Jakoubek, and
A. L. Schmeltekopf, Observations of the nighttime abundance
of OC10 in the winter stratosphere above Thule, Greenland,
Science, 4, 550-555, 1988.
Toohey, D. W., J. G. Anderson, W. H. Brune and K. R. Chan, In
situ measurements of BrO in the Arctic stratosphere, Geophys.
Res. Left., 17, 513-516, 1990.
Wahner, A., J. Callies, H. P. Dom, U. Platt, and C. Schiller, Near
UV atmospheric absorption measurements of column abun-
dances during Airborne Arctic Stratospheric Expedition, Jan-
uary- February 1989, 3, BrO Observations, Geophys. Res.
Left., 17, 517-520, 1990.
World Meteorological Organisation/UNEP, Scientific assessment
of ozone depletion, Global Ozone Research and Monitoring
Project, Rep. 25, Geneva, 1991.
D. Cariolle and P. Simon, M6t6o France, Centre National de
Recherches Mtorologiques, 42 Ave. Coriolis, 31057 Toulouse,
France.
M. Chipperfield and D. J. Lary, Centre for Atmospheric Sci-
ence, Department of Chemistry, University of Cambridge, Lens-
field Rd., Cambridge, CB2 1EW, U.K.
R. Ramaroson, Office National d'Etudes et de Recherches
Arospatiales, 92322 Chtillon, France.
(Received December 30, 1991;
revised December 15, 1992;
accepted December 15, 1992)