Q. 1. R. Mereorol. SOC. (1994). 120, pp. 453-482 551.510.413.2:551.513.1
A three-dimensional model study of nitrogen oxides in the stratosphere
By D. J . LARY, J . A . PYLE* and G. CARVER
University of Cambridge, UK
(Received 25 November 1992; revised 29 July 1993)
SUMMARY
A new three-dimensional model for stratospheric chemistry studies is described. The model, a spectral
general circulation model, is developed from the model used at the European Centre for Medium-range
Weather Forecasts. Schemes to describe atmospheric Chemistry and photochemistry as well as advection of
trace gases have been implemented. In this study a perturbed period in January 1987 is studied. The model
performs well. An interesting feature, similar to the Noxon cliff, is reproduced in which the NO2 column is
reduced significantly in high latitudes. In this calculation the cliff is produced by gas phase conversion of NO2
to N2OS.
1. INTRODUCTION
It was first suggested in the early 1970s that stratospheric ozone could be depleted
by a variety of man-made compounds (Johnston 1971; Crutzen 1971; Molina and Rowland
1974). Projections of the likely ozone loss were based first on one-dimensional photo-
chemical models, which consider only the variation in species concentration with altitude
(see the references above), and subsequently on two-dimensional (2-D) models (e.g.
Pyle 1980). The latter models include transport by the mean meridional circulation, but
eddy transport, associated with large-scale atmospheric waves, must be parametrized.
Some 2-D models have included the feedback between ozone amount, radiative heating
and the meridional circulation (e.g. Harwood and Pyle 1975), but many have not. In
either case there is a fundamental limitation in describing a three-dimensional (3-D)
system with only two spatial variables (see, for example, Murgatroyd (1982) for a detailed
discussion).
The years since ozone loss was first hypothesized have seen the atmospheric chemistry
community become more aware of the importance of an adequate description of 3-D
atmospheric transport. With increases in computing power a number of atmospheric
chemistry studies that use 3-D models have been performed. Early studies included
Mahlman and Moxim (1978) who introduced a very simple ‘ozone-like’ tracer into
their general circulation model (GCM). Cariolle and Deque (1986) developed a highly
parametrized ozone scheme for their model, which produced a very satisfactory ozone
distribution. Subsequently, other stratospheric chemistry schemes of varying sophis-
tications have been included into 3-D models (e.g. Grose et al. 1987; Kaye and Rood
1989; Rood et al. 1990; Granier and Brasseur 1991; Austin and Butchart 1992).
The steady increase in computer power has now led to a situation where integrations
of stratospheric chemistry schemes in 3-D models are quite feasible for process studies.
On the other hand, integrations covering many decades for assessment studies are not
yet practical unless run at very low resolution, but such integrations can be expected
before long, especially as the importance for climate change of ozone modification in the
lower stratosphere is now recognized (see, for example, Lacis et al. 1990).
This paper describes the first results with a new 3-D stratospheric chemistry model,
based on a GCM from the European Centre for Medium-range Weather Forecasts
(ECMWF), which is being developed as a climate/atmospheric chemistry research tool
* Corresponding author: Centre for Atmospheric Science, Department of Chemistry, University of Cambridge,
Lensfield Road, Cambridge CB2 IEW, UK.
453

454 D. J . LARY, J. A. PYLE and G. CARVER
by the UK Universities Global Atmospheric Modelling Programme (UGAMP). Various
modifications to the model have been effected to allow the inclusion of atmospheric
chemistry. These are described in section 2, where the model is discussed in more detail.
Section 3 contains a description of the chemistry and photochemistry schemes included
for this integration. The model described here includes detailed treatments of the
chemistry that involve ozone and the nitrogen-, hydrogen- and chlorine-oxide families.
A description of the overall performance of the chemical model in a two-week integration
initialized dynamically in midJanuary 1987 is given in section 4. Emphasis is given to an
analysis of the nitrogen budget in the model. A feature resembling the Noxon cliff, in
which there is a sharp decrease in the NOz column as the pole is approached, is found;
this is dealt with in more detail and its causes discussed in section 5. Our conclusions are
given in section 6.
2. THE UK UGAMP GENERAL CIRCULATION MODEL
The UK UGAMP GCM (UGCM) is a full 3-D primitive-equation model, based on
the ECMWF cycle 27 GCM (Tibaldi et al. 1990). The model includes a detailed set of
parametrizations for unresolved dynamical and physical processes. The UGCM differs
from the ECMWF cycle 27 version mainly by the inclusion of a facility for tracer advection
and a detailed stratospheric chemistry scheme, described below. The model was run on
the Cray XMP/48 at the Atlas Computer Centre at the Rutherford Appleton Laboratory.
The primitive equations are expressed in terms of vorticity, divergence, temperature,
specific humidity and log surface pressure. The spectral technique (Hoskins and Simmons
1975) is used to solve the equations in the horizontal, whilst a finite-difference scheme
is used in the vertical. For this study the UGCM was run with a triangular truncation of
the retained modes in wave-number space up to zonal wave number 21 (usually written
as T21). This corresponds approximately to a grid of 6" X 6" in latitude and longitude.
Physical parametrizations are calculated in grid-point space and then transformed to
spectral space. The hybrid sigma-pressure vertical coordinate of Simmonds and Burridge
(1981) is used, in which the lowest model levels are sigma levels whilst the uppermost
are pressure levels. The model has 19 levels in the vertical, from the surface up to 10 mb.
The resolution in the stratosphere is relatively coarse, with levels at approximately
10,30, SO, 75,100 and 140 mb. The earth's orography is spectrally fitted and smoothed
before being included as a boundary condition in the model.
The semi-implicit time integration scheme of Hoskins and Simmons (1975) with an
Asselin (1982) time filter is used. At a horizontal resolution of T21 the dynamical time
step would normally be 45 minutes, but stable integration of the chemical scheme
(forward Euler) required a 5-minute time step. In this study, therefore, a S-minute step
was used for the integration of both chemistry and dynamics. Our more recent work has
used an implicit scheme for the chemistry, allowing longer time steps. Solution of the
radiative-transfer equation to compute the radiative fluxes is performed once every 3
hours on a reduced grid to save computer time. All other physics parametrizations
(including radiative heating rates) are calculated at each step on the full model grid.
Horizontal diffusion is applied in spectral space to all the model's prognostic
variables, using a sixth-order operator with an &hour e-folding time at wave number 21.
Horizontal diffusion is increased for the top five model levels to provide a 'sponge-layer'
with a 15-minute e-folding time at the top level for wave number 21. This is necessary
to control the large wind speeds that would otherwise develop. It has the consequence
of removing small-scale features, but the large-scale features are unaffected. Vertical
diffusion, based on Monin-Obukhov similarity theory, is also applied in the planetary

3-D STRATOSPHERIC CHEMISTRY MODEL 455
boundary layer and in the free atmosphere dependent on the stability. The radiation
parametrization includes short- and long-wave fluxes in clear and cloudy sky conditions.
It allows for partial cloud cover in any model layer and includes multiple scattering
(Geleyn and Hollingsworth 1979; Geleyn etal. 1982). A separate radiative code, described
later, is used for the photolysis-rate calculations. Convection is modelled using a modified
version of the Kuo (1974) scheme. Shallow convection is treated as part of the model’s
vertical diffusion (Tiedtke 1983).
For the UGAMP research a tracer advection facility was incorporated in the model.
This allows passive or chemically active tracers to be advected in the model, using the
model winds updated on each time step. This ‘on-line’ calculation allows, for example, the
possibility of incorporating the modelled species fields into the radiation parametrization,
although this was not done for this study. The tracer advection code was written so that
the maximum number of tracers is limited only by the available computer memory. In
practice more than 30 tracers becomes prohibitive.
The tracer advection is discretized, using the spectral method in the horizontal and
the finite-difference scheme in the vertical as for the other prognostic variables. However,
it is well known that the spectral method suffers deficiencies when representing fields
with large horizontal gradients over a few grid points (Rood 1987). These are often
highlighted by the occurrence of negative mixing-ratio values and regions of undershoot
and overshoot where the minimum or maximum mixing-ratio value is exceeded, although
conservation globally is ensured. There are two main causes of these problems. First,
the grid associated with a spectral truncation supports more waves than is actually used
by the model; thus small-scale information on the grid can be lost. Second, Gibbs
phenomena can occur when discontinuities develop in the field; this has important
consequences for chemical source regions. For distributions of chemical species in which
sharp gradients form (such as those with a pronounced diurnal cycle) the spectral
technique may not be the most appropriate, and numerical errors will arise.
The vertical advection scheme can also produce undershoot and overshoot since it
is first-order accurate with the model’s variable vertical spacing. The vertical advection
scheme can also give rise to regions of negative mixing-ratio values.
We have taken a very crude approach in these first calculations to the problem of
negatives by simply setting negative values to zero. This, of course, leads to non-
conservation, which would be very serious in a long integration but is acceptable in the
short run presented here. Our recent model developments have included an improved
scheme for vertical advection (Thuburn 1993), which significantly reduces the problem,
as well as a borrowing scheme in the chemistry that ensures conservation.
3. STRATOSPHERIC PHOTOCHEMICAL SCHEME
(a) The chemistry scheme
A comprehensive photochemical package containing 74 chemical and photochemical
reactions is used to describe the stratospheric chemistry of Ox, NO,, HO, and CIO,
(Table 1). The rate constants were taken from DeMore et a f . (1987). The rapid time-
scales on which certain photochemical processes occur can impose a severe limitation on
the time step that must be used in chemical models if numerical stability is to be
maintained. A well proven technique, which provides a computational solution to this
as well as a conceptual simplification, is the family approach, in which continuity equations
are solved for a family, and not for species individually. A family is a group of constituents
that are in photochemical equilibrium with each other. The interchange between family

D. J. LARY, J. A. PYLE and G. CARVER
TABLE 1. REACTIONS CONSIDERED
0 + 0 2 : 0 3
0 + 03+ 20z
o(~D) + N~ + 00~) + N~
O('D) + 02-r o(3~) + o2
O('D) + HzO + 20H
OH + O + H + 02
O2 + H 5 HOZ
OH + 0,- H02 + O2
H + 03+ OH + Oz
OH + OH+H20 + 0
HOZ + O + OH + 0 2
OH + H02+ H20 + 0 2
NO2 + O+ NO + 0 2
NO + O,+ NO2 + O2
HNO, + OH + NO3 + H20
NzO + O('D) + 2N0
NO2 + 0 3 + NO3 + 02
NO2 + OH 5 HNO3
H02 + HOz + H202 + 0 2
HZ02 + OH+HzO + HOz
OH + CH4 + CH3 + H20
O('D) + CH4 + CH3 + OH
NO + H02+ NO2 + OH
H02 + O3-r OH + 202
H02 + NOz 5 HOzNOz
H02NOz 5 HO;! + NO2
CFCI, + O('D) + 3CI
C1 + O3 + CIO + O2
CIO + NO+ C1 + NOz
H02N02 + OH + HzO + 0 2 + NO2
CF2C12 + O('D) + 2C1
CIO + 0 + CI + o2
CH, + C1+ CH3 + HCI
H2 + CI + H + HCI
HO2 + C1+ 0 2 + HCI
OH + HCI+ H20 + CI
CIO + NO2 5 C1ONO2
C10N02 + 0 + products
OH + HOCl + H20 + C10
H02 + CIO + HOCl + 0 2
0 + HOCl + OH + CIO
N + NO+ N2 + 0
N + 02' NO + 0
CC4 + O('D) + 4CI
CH3CC13 + O('D) + 3CI
CH3CC13 + OH + 3C1
CH3CI + O('D) + CI
CH&I + OH + CI
CHFZCI + O('D) + CI
CHF2CI + OH + CI
CFzCICFCI2 + O('D) + 3CI
NO3 + NO2 3 NzOS
N205 2 NO3 + NO2
c10 + CIO 5 c1202
Cl2OZ 3 c10 + ClO
o2 + hv+ 0 + 0
0, + hv + oz + O(-'P)
NO,? + hv+ NO + o(3~)
0, + hv-r Oz + O('D)
NO + hv+ N + 0
NO3 + hv+ NO + O2
NO3 + hv-r NOz + 0
HNO, + hv+ NO2 + OH
NzOS + hv+ NOz + NO3
HOzNOz + hv+ NO2 + €302
N20 + hv+ N2 + O('D)
C1ONOz + hv+ CI + NO3
CCl, + hv+ 4CI
CFCl3 + hv + 3C1
CFZCl2 + hv+ 2CI
CHFZCl + hv + CI
CF2CICFCI2 + hv + 3CI
HOCl+ hv+ CI + OH
ClZO2 + hv+ c102 + CI + 2c1 + o2
members occurs in a time-scale that is much less than the net production/loss time-scale
of the family as a whole. Since the photochemical lifetime of the family is greater than
any of its individual members, a computational advantage results, because numerical
stability can be maintained with much longer model time steps than would otherwise be
possible. Furthermore, since the role of atmospheric transport is generally easier to
identify in the global distribution of the family, rather than for individual family members,
a conceptual simplification occurs.
A careful choice of family members is very important since, if the assumption of
photochemical equilibrium is not valid over the whole model domain, quite large errors
can be incurred (e.g. Douglass et al. 1989; Austin 1991). This problem must be avoided
by a careful analysis of the lifetimes of the constituents involved. Generally, if the lifetime
of a given constituent exceeds a few hours, caution should be exercised in its inclusion

3-D STRATOSPHERIC CHEMISTRY MODEL 457
in a family. For example, photolysis of N205 can be rapid at low latitudes in the
stratosphere; N2O5 has a short lifetime and could be included in a family along with NO
and NO2. However, N205 photolysis is slow (or zero) at high latitudes during winter, the
lifetime of N 2 0 5 increases and so an assumption of immediate photochemical equilibrium
with NO and NO2 is no longer valid.
An additional factor in deciding the family members is the relative importance of
the particular constituents. For example, in some circumstances up to 50% of the total
reactive nitrogen can be in the form of N205, so that a careful treatment of N205 is
essential. However, another constituent of comparable lifetime to N2O5 is H02N02, but
H02N02 is present at much lower concentrations (typically < 1 % of the nitrogen species)
so that a slight departure from photochemical equilibrium for H02N02 would not have
such a large potential impact.
An analysis of stratospheric lifetimes was conducted and this led to the choice of
families given in Table 2. With two exceptions the families listed in Table 2 are transported
by the model winds as described in the previous section. First, as in the Cambridge 2-D
model, HO, is assumed to be in photochemical equilibrium, and is not transported by
the model winds. This is a good approximation, since HO, has a very short lifetime in
the troposphere and stratosphere. If this assumption is not made, then the time step that
would be required for numerical stability (< 1 minute) is such that the model could not
be used for integrations of much longer than one day. Second, in the short integration
described here, the long-lived source gases are not transported but are simply specified.
The continuity equation is solved for each family. For example,
where the photochemical production rate of 0, is denoted by Vi, the chemical destruc-
tion rate of 0, is denoted by ZLi [O,], and r is the wind vector. Once this equation is
solved for the concentration of Ox, the concentration of the family members (03, O('D)
and OCP)) are calculated by using the family ratios
where j is the photolysis rate, k the thermal reaction, M stands for any third body
involved in a reaction and the square brackets denote concentrations.
A more detailed description of the model is given by Lary (1991). The kinetic data
are taken from DeMore et al. (1990), apart from the following: the absorption cross-
section for the Herzberg continuum of molecular oxygen is taken from Nicolet and
Kennes (1986) and WMO (1986); the absorption cross-section for the Schumann-Runge
bands of molecular oxygen are calculated using the parametrization of J. E. Frederick
(1985, private communication) (see also WMO 1986); the temperature dependence of
the O3 absorption cross-section in the spectral region 264 nm < A < 345 nm is calculated
using a quadratic fit to the dataset of A. M. Bass presented by J. E. Frederick (1985,
private communication) (see also WMO 1986); the absorption cross-sections of NO in

458 D. J. LARY, J. A. PYLE and G . CARVER
the d(0-0) and d(1-0) bands are calculated using the parametrization of Allen and
Frederick (1982). This parametrization applies for the region above 20 km, and for solar
zenith angles up to 85". Outside this domain the parametrization does not apply, and the
absorption cross-section is set to zero. The temperature-dependent absorption cross-
sections of the halocarbons CH,Cl, CC14, F1 1(CFC13), F12(CF2C12), and F22(CHF2C1)
are calculated using the parametrizations of Simon et al. (1988) (although the impact of
this photolysis is of course negligible in the very short integration discussed below).
The stratospheric chemistry scheme is being continually developed and updated.
Our most recent studies include yet more detailed chemical packages, for example the
chemistry of the bromine compounds. Nevertheless, the scheme presented here is quite
sufficient for the detailed study of the nitrogen oxides presented below.
(b ) Radiative transfer model
The radiative transfer model used in this study to calculate atmospheric photolysis
rates is a new implementation of the scheme described by Meier et al. (1982). It has been
extended after Anderson (1983) to describe correctly the radiation field for solar zenith
angles greater than 75". The radiation into any volume element of the model atmosphere
has four contributions: (A) the direct solar flux, (B) the diffuse flux incident from all
directions, (C) the ground reflection of the direct solar flux, and (D) the ground reflection
of the diffuse flux. This is illustrated schematically in Figure 1.
The radiation field is calculated by solving the integral equation of radiative transfer.
The detailed mathematical description of the four contributions is described by Meier et
al. (1982) and Lary (1991). The direct flux is treated by using a full spherical geometry,
and the scattered flux by the plane-parallel approximation. Using the plane-parallel
approximation to describe the multiple scattering, results in an underestimate of the
radiation field for solar zenith angles greater than 93". We have usually carried calculations
up to 96", with some loss of accuracy at the largest angles. The accuracy of the method
has been demonstrated by Anderson (1983). Our implementation has been justified by
a number of studies that compare the chemistry/radiative model with atmospheric
measurements (Lary 1991; Lary et al. 1991; Lary and Pyle 1991a, b). The studies presented
in this paper assume clear sky conditions.
Photolysis rates are calculated by making use of an enhancement factor, or nor-
malized source function, SA (Meier et al. 1982), defined as the total number of photons,
cos 121
Surface
Figure 1. Schematic diagram of the radiative-transfer model used for the calculation of photolysis rates.
(Adapted from Meier et al. (1982) and reproduced from Lary and Pyle (1991a) with the permission of the
Journal of Atmospheric Chemistry).

3-D STRATOSPHERIC CHEMISTRY MODEL 459
FA, integrated over all directions, which is available for photolysis at any given point in
the atmosphere, normalized by the number of photons incident at the top of the
atmosphere, FoA,
Any photolysis rate j , can readily be calculated from a knowledge of the solar flux
incident at the top of the atmosphere, F,,, the absorption cross-section, q, the quantum
efficiency, @A, and the enhancement factor, SA, using
Note that the enhancement factor SA is a function of wavelength A, solar zenith angle x,
altitude z , and ground albedo Aground. SA also depends on which ozone, temperature and
aerosol profiles are used. In this study atmospheric aerosols have not been included, and
a single ozone and temperature profile was used in the calculations of the photolysis
rates. Studies conducted more recently have used the local profiles.
4. RESULTS
For the initial run with the model described above we have carried out a study of
the evolution of the dynamical and chemical fields in a 14-day integration during a
disturbed northern hemisphere winter.
The model was initialized dynamically with the ECMWF analysis for ~ ~ G M T 15
January 1987 and integrated for two weeks. There are no global datasets available for
this period to initialize the chemistry. We have, therefore, taken the simplest possible
approach. The chemical fields were initialized from the Cambridge 2-D model. Thus,
initially, the chemical fields were zonally symmetric, an unrealistic condition and a point
discussed further later.
( a ) Dynamical fields
The stratosphere of January 1987 was perturbed by an extremely intense mid-winter
warming that developed early in the month and persisted well into February. The details
of the warming are described by Naujokat et al. (1987). During mid to late January the
main features changed little: a pronounced wave-number one pattern at 10 mb, with a
corresponding geopotential height minimum of about 28 km centred over northern
Scandinavia and Russia, moved only very slowly to the south-west during this period.
Figure 2 shows the 10mb geopotential maps for 19, 21 and 27 January derived from
Stratospheric Sounding Unit (SSU) data. Temperature fields are shown in Fig. 3.
Temperatures at the pole are high in late January, as shown in the temperature map for
19 January at 10 mb (Fig. 3(b)). Notice the very strong temperature gradients with very
low temperatures associated with the vortex over northern Scandinavia.
The model dynamics were initialized during the warming; we can, therefore, make
no statement about the prediction of the onset of the perturbation. The model, even at
the low resolution used, followed the observed features reasonably well, especially during
the first week of the integration. (Similarly good agreement over about the first week of
integration was found in studies of the 1991/92 winter, see Carver et al. (1994).) Figure
4 shows the calculated 10mb geopotential heights for 19, 21 and 27 January. The
geopotential minimum on 19 January was about 28.6 km, slightly higher than inferred

460 D. J. LARY, J . A. PYLE and G . CARVER
Figure 2. Geopotential heights (dagpm) for the northern hemisphere 10mb surface from SSU data for: (a)
19 January 1987, (b) 21 January 1987, and (c) 27 January 1987. (Courtesy of Drs A. O’Neill and M. Bailey,
Meteorological Office).

9oW
9 0 h
3-D STRATOSPHERIC CHEMISTRY MODEL
C)
I
461
3 E
Figure 3. As Fig. 2 but for temperatures (K).

462
9OW
9 O h
9 0 E
9OE

3-D STRATOSPHERIC CHEMISTRY MODEL 463
Figure 4. Calculated geopotential heights for the northern hemisphere 10 mb surface for: (a) 19 January 1987,
(b) 21 January 1987 and (c) 27 January 1987. Contours every 200 kgpm.

464 D. J. LARY, J . A . PYLE and G . CARVER
from SSU data. The position and orientation of the modelled vortex agree quite well
with the observations. On 21 January the agreement is even better, with the position
and magnitude of the observed and modelled geopotential minimum coinciding almost
exactly. By 27 January, at the centre of the vortex, the geopotential height was about
28.6 km, lower than observed. The modelled vortex extends much further across the
Greenwich meridian than observed (Fig. 2(c)) and the high-pressure system is not
reproduced particularly well. Nevertheless, the model retains the main feature of the
observations, with a vortex centred over the European sector.
The model 10 mb temperatures on 19,21 and 27 January are shown in Fig. 5. Notice
that the lowest temperatures in the sequence are found on 19 January, being less than
200 K, and that on both 19 and 21 January high temperatures are observed over the pole.
By 27 January the lowest temperature had increased to nearly 210 K, with the temperature
minimum moving south and west during the integration. These features are qualitatively
similar to those shown in the observed temperature field (see Fig. 3), although by 27
January the observed structure looks quantitively quite different to that observed. In
particular, the high temperatures observed close to the pole are considerably under-
estimated in the model integration.
Our objective in this paper is not to discuss in detail the dynamical behaviour of the
model during this particular period up to the end of January 1987. Instead we want to
discuss some aspects of the chemical behaviour of the model during a disturbed winter
period. For this purpose it is not, of course, necessary that the integration bears
any resemblance to the observed circulation, particularly as there are few chemical
observations against which to compare the model during this same period. Nevertheless,
it is clear from the above that the dynamical model did indeed perform quite well in a
predictive role during this period.

3-D STRATOSPHERIC CHEMISTRY MODEL
Figure 5. As Fig. 4 but for calculated temperature (K).
465

466 D. J. LARY, J . A . PYLE and G . CARVER
Figure 5. Continued
(b) Total ozone
Figure 6 shows the calculated vertical ozone column for 15, 17 and 19 January.
Notice that the largest contribution to the ozone column comes in the lower stratosphere,
where the vertical resolution of the model is relatively poor.
The model begins with a zonally symmetric ozone field, but very quickly a com-
plicated structure is developed and by 19 January the maximum values form an H-shape
stretching from the North American continent across the pole to central Russia. On this
day much lower values of ozone are found over the United Kingdom and Scandinavia.
Figure 7 shows a Total Ozone Measuring System (TOMS) map for 19 January 1987.
There is excellent qualitative and quantitative agreement between these observations
and the model results (Fig. 6(c)). For example, the model maximum of nearly 500
Dobson units (DU) at 60"N, 120"E can be compared with a TOMS maximum also of 500
DU in the same region. Such good agreement may seem surprising at first sight, since
the initially zonally symmetric conditions were quite unlike the atmospheric state.
However, the ozone column is determined largely by atmospheric dynamics. For example,
the inverse correlation between total ozone and tropopause height has been known for
many years (Dobson et al. 1927). (Vaughan and Price (1991) have recently pointed to
an even better correlation between total ozone and the vorticity around the tropopause
level.) The good comparison between model and data arises simply from a realistic
dynamical representation of the lower stratosphere soon after initialization with the
correct meteorological analysis. This is confirmed by Fig. 8 that shows the model 100 mb
geopotential height for 19 January. Notice the extremely good correspondence between
the orientation across the pole of the geopotential minimum and the ozone column (Figs.

3-D STRATOSPHERIC CHEMISTRY MODEL
180
461
Figure 6. The northern hemisphere total ozone column (DU) for: (a) 1400 GMT 15 January 1987, (b) 1200 GMT
17 January 1987, (c) 1200 GMT 19 January 1987.

468 D. J . LARY, J. A. PYLE and G . CARVER
Figure 6. Continued.
,
Figure 7. The northern hemisphere total ozone column as observed by the TOMS instrument on 19 January
1987.

469 3-D STRATOSPHERIC CHEMISTRY MODEL
Figure 8. Modelled geopotential height for the northern hemisphere 100 mb surface on 19 January 1987.
Contours every 100 kgpm.
6(c) and 7). Notice also that the arms of the H-shape in the modelled total ozone
correspond to the geopotential minimum over North America and Siberia.
At later dates in the integration, detailed comparison between TOMS observations
and the model become less good. The main features of ozone maximum and minimum
values are still common to model and observation but the precise positions of the features
move apart. Clearly, in the lower stratosphere, the calculated (forecast) dynamical fields
drift away from the analyses, leading to a less realistic total-ozone field when compared
with observations. Nevertheless, the column-ozone results show that even with a simple
chemical initialization realistic composition fields can soon be established in the model,
as long as in this case the simple initialization captures the observed zonal mean
structure of the atmosphere. For example, in this integration the ozone field agreed with
observations by the third day.
Our more recent model developments have included an improved initialization in
which the initial 2-D constituent fields are transformed into an 'equivalent potential
vorticity latitude-potential temperature' coordinate system (Lary et al. 1994, private
communication), which should lead to significant improvements. Nevertheless, for the
purposes of this paper, the zonally symmetric initialization is perfectly acceptable.
(c) O3 mixing ratios at 10mb
Figure 9 shows the development of the ozone field at 10 mb during the integration.
Figure 9(b), 12 hours after the start of the run, shows that the strong cross-polar flow
has pushed the ozone minimum well off the pole, towards 60"N, MOW. Thirty-six hours
later (Fig. 9(c)) a horseshoe shape has been established, quite unlike the initial conditions.

470 D. J. LARY, J. A. PYLE and G . CARVER
180
Figure 9. Modelled ozone mixing ratio (p.p.m.v.) at 10 mb in the northern hemisphere for: (a) 1400 GMT 15
January 1987, (b) oo00 GMT 16 January 1987, (c) 1200 GMT 17 January 1987 and (d) 1200 GMT 21 January 1987.

471
Figure 9. Continued.

412 D. J. LARY. J. A. PYLE and G . CARVER
By 21 January (Fig. 9(d)) the original minimum has been split into two, with one centred
over the vortex at 60"N, 40"E and the secondary minimum being close to the Aleutian
anticyclone. (Calculations with the improved chemical initialization do not show this
second minimum.) In between the minima lies a region of higher mixing ratios over the
pole, air that has been advected from lower latitudes.
Figure 9(c) is reminiscent of the Limb Infra-red Monitor of the Stratosphere (LIMS)
ozone fields during January and February 1979 studied by Leovy et al. (1985). Figure 10,
based on their paper, traces with a dashed line the tongue of low-latitude air, poor in
ozone, that has been pulled towards the North Pole at 10mb. It seems clear that the
features shown in Figs. 9 and 10 are similar.
( d ) Column NO2
The vertical column of NOz from 15 January is shown in Fig. 11. Initialization from
the 2-D model leads to a zonally symmetric field for the odd-nitrogen family (NO, in
Table 2). Figure 11 shows a highly non-zonal field for NO2 that arises because the rapid
chemical partitioning within the odd nitrogen family varies with zenith angle. Thus a
dominant feature is the terminator that crosses the lower (Atlantic) side of the map. In
the sunlit portion of the globe, NO2 levels are generally low while night-time values are
high, even over the pole.
Figure 12(a) shows the same field a few days later on 19 January. Again the variation
across the terminator stands out strongly. Following, for example, the 45"N latitude circle
anticlockwise from dawn to dusk, note the general increase in NO2 associated with its
Figure 10. Ozone mixing ratio (p.p.m.v.) on the northern hemisphere 10 mb surface on 6 February 1979 as
measured by the Nimbus 7 Limb Infra-red Monitor of the Stratosphere (LIMS) instrument. (Figure produced
by J . C. Farman for Phil. Trans. R . , SOC. London A, 323 after Leovy er al. 1985, and reproduced with

3-D STRATOSPHERIC CHEMISTRY MODEL 473
Figure 11 . Calculated NO2 column (10’’ molecules cW2) for the northern hemisphere, 1200
1987. The figure shows the initial conditions, before integration.
TABLE 2. SPECIES GROUPINGS USED IN THIS STUDY
’ GMT 15 January
F d e S
0, = O(’D) + OcP) + O3
NO,= N + NO + NO2 + NO3
HO,= H + OH + H02 CIO, = c1 + CIO + c 1 2 0 2
Reservoir species
N205, HN03, H02N02 , CION02, HCI, HOCl
Source gases
N20, H20, CH4
F11 (CFCII), F12 (CF2Cl,), F22 (CHF2Cl)
F113 (CF2 CICFC12), CHjCI, CH3CC13, CCI4
release by photolysis of N205. Similarly, notice the night-time decay of NO2 at the same
latitude: just after sunset, column NO2 amounts are at their highest and gradually decay
during the night as N205 is produced. The mid-latitude behaviour of the model is thus
in excellent agreement with the observed NO2 diurnal variation (see Webster et af. 1990;
Lary et af. 1991).
Figure 12(a) shows a very interesting feature, not shown in Fig. 11, when the high-
latitude column NO2 amount drops to very low values (e.g. around WON, 9OOW). This
feature is reminiscent of the Noxon cliff, a feature first measured by the late John Noxon
(Noxon 1979; Noxon et af. 1979, 1983), in which the column NO2 can drop dramatically
with increasing latitude. Figure 12(b) shows the calculated NOz column for 27 January.
A Noxon cliff-like feature is again evident, although in this case the lowest values are
found over the pole and not displaced to the west as found in Fig. 12(a). Later we will
return to a detailed discussion of this behaviour of NO2 in high latitudes, after first
presenting results showing the distribution of the two major nitrogen reservoirs, N205
and HN03.

414 D. J. LARY, J. A. PYLE and G . CARVER
(a) The No, mumn decrBBS8S as the equatw is approached due to a deaease in total Q.
High levels of No, associated with sunset \
levels of N02due to low temperatures
and low leyels of ozone. North Amxica and Greenland by the Low values Of No2
stroong a m plar pt associated with sunrise
\- Cisplaced polar vortex with low Low kwals of bQ, associated win polar night air displaced towards
o jet
I80
(b) -
The low levels of NO,associated with polar
night are now centered on the north pole
since the strength of the moss polar jet has
subsidej.
considerably since January 19 and so
m e reactive nitrogen is in the fwm of NO,within
the polar vortex rqion ,
Figure 12. The modelled NO2 column molecules cm-*) for the northern hemisphere for: (a) 1 2 0 0 ~ ~ ~
19 January 1987 and (b) 1200 GMT 27 January 1987. Note: In (a) the sharp gradients in NOz are shifted to lower
latitudes owing to the presence of a strong cross-polar jet.

3-D STRATOSPHERIC CHEMISTRY MODEL 475
( e ) N2°5
Figure 13 shows the N20s mixing-ratio fields at 10mb for 19 and 25 January. The
N205 values are highest in polar regions. When the NO2 minimum is displaced off the
pole on 19 January, so is the N2O5 maximum. On 25 January maximum Nz05 mixing
ratios and minimum NO2 column (not shown) again coincide, this time over the pole.
Like NO2, N205 has a diurnal variation. However, while the NOz photochemical
time constant is very short, being on the order of minutes, that for NzO5 is longer and
on the order of hours. Thus, N205 does not show a strong variation across the terminator.
Instead minimum values are found in the late afternoon and into the early part of the
night (easily seen following the 30"N latitude circle in Fig. 13(a) when the lowest N20s
is found near 80"E). Minimum values occur following the relatively slow photolysis during
the day and before substantial night-time production has taken place. Similarly, large
concentrations are found at the end of the night and during the first few hours of sunlight.
Figure 13. The modelled N205 mixing ratio (p.p.b.v.) at 10mb in the northern hemisphere for: (a) 1200 GMT
19 January 1987 and (b) 1200 GMT 25 January 1987.

476 D. J. LARY, J. A. PYLE and G. CARVER
(f) HNO3
The ratio of the concentrations of nitric acid to NO, (defined as the sum of all odd-
nitrogen compounds) at 10mb on 21 January is shown in Fig. 14. There is no strong
correlation of HN03 with the NO2 field for the same day. Instead the nitric acid behaviour
resembles that of 03. It is evident that the HN03 is influenced after six days of calculation
primarily by the initialization and subsequent dynamics, and that chemistry has had only
a minor influence on its distribution. However, there is some correspondence between
the HN03 concentration and the temperature structure. In particular, low HN03, relative
to NO,, is found in the warm cross-polar jet.
Low concentrations of W 3 are associated
with the warm moss polar jet.
\ A local M 3 minim due to
The cr& polar / Jetentram.
a warm pool of air brought
to high latitudes by the
cross polar jet
\ I
The highest concentrations of W are associated
with the very cold polar vortex
Figure 14. The modelled ratio (%) of HN03 to NO, (NO, + HNO, + 2N,05) at 10mb in the northern
hemisphere for 21 January 1987.
5 . DISCUSSION
The Noxon cliff was discovered in the late 1970s when a series of measurements of
the NOz column, made at various times and latitudes, revealed a hitherto unexpected
feature (Noxon et al. 1979,1983). Noxon found that in the winter and spring the generally
increasing trend of the NO2 column with latitude could on occasion be dramatically
reversed. The NO2 column could then drop by a large factor from perhaps 7 x 10’’
molecules cm-z to less than 1015 molecules cm-z, over a few degrees of latitude.
It is clear that the normal partitioning within the nitrogen family is perturbed during
these situations. A number of possibilities exist. Firstly, the NO2 could have been
converted to N 2 0 5 by the reactions
NOz + 03+ NO3 + O2
NO3 + NO2 + M + N205 + M
where production of N 2 0 5 is rate limited by the reaction between NOz and 03. This
reaction is part of the normal diurnal cycle of NOz and would be favoured by long periods
of darkness in the polar winter. Slow photolysis of Nz05, the main destruction of N205,

3-D STRATOSPHERIC CHEMISTRY MODEL 477
in the cold high-latitude stratosphere could then keep the NO2 low as the air parcel is
advected from high latitudes to the measuring sites. This idea was put forward by,
amongst others, Solomon and Garcia (1983a,b) and Callis et al. (1983). Solomon and
Garcia (1983a) suggested in particular that the details of the cliff might depend strongly
on the temperature dependence of N205 photolysis.
Another possibility suggested by Noxon (1979) is that HN03 is the reservoir. HN03
is produced by the reaction
NO2 + OH + M + HN03 + M.
This reaction is favoured by low temperatures and requires sunlight to produce the O H
radical. HN03 could also be the reservoir if heterogeneous reactions are involved. For
example,
N2Os(g) + H2O(s)- 2HNO36)
is known to be effective both on nitric acid trihydrate particles (polar stratospheric
clouds) and aerosol surfaces (Hanson and Ravishankara 1991). The reaction on polar
stratospheric clouds obviously needs the low temperatures required for the formation
of these surfaces. The reaction on sulphate aerosol appears to be independent of
temperature.
The run considered above includes only gas-phase chemistry so we are unable to
say anything about the role of heterogeneous chemistry. However, it is clear that, on the
basis of the calculations described in section 4, gas-phase formation of N205 leads to the
modelled Noxon cliff. This is amply demonstrated in Figs. 12 and 13 where there is a
very strong anticorrelation in high latitude between NO2 and N20S. The large area of
low NO2 corresponds exactly to the region of highest N205 concentration. Furthermore,
this is also where the ratio of H N 0 3 to NO, becomes low (Fig. 14). Whether or not
HNO, would subsequently be formed on sulphate aerosol-temperatures are too high
for polar stratospheric clouds to form-it is clear that, under the dynamical conditions
of January 1987, formation of N20s is the precondition for the formation of the Noxon
cliff.
To consider the situation in more detail, consider the rate-limiting step for formation
of N205
NO2 + 03+ NO3 + O2
where T is temperature, and for its destruction
k = 1.2 x exp (- 2450/T)cm3s-'
N205 + hv+ NO2 + NO3.
It is clear from these two reactions that high concentrations of N205 are favoured by
darkness, when the photolysis of N205 is zero, and by both high concentrations of ozone
and high temperatures, which lead to enhanced rates of production of N205. All of these
conditions are present over the pole in January. One of the major features of the model
simulation is a warm cross-polar jet which brings air, rich in ozone, from lower latitudes
across the pole and around the edge of the vortex. This is very evident in, for example,
Figs. 5(b) and 9(d) which show high temperatures and ozone, respectively, at 10 mb over
the pole on 21 January. At the same time, the flow across the pole leads to perhaps 18
hours of darkness for individual air parcels, as revealed by tracer experiments in the
model. Under these conditions most of the NO2 can be removed, as shown schematically
in Fig. 15.
The exact position of the cliff can also be explained. On 19 January the cross-polar
flow around the vortex was strong, and the low NO2 was advected away from the pole

478 D. J . LARY, J. A . PYLE and G. CARVER
Parcel which has beer) moved into darknes's
Last 6rccUliET
After 18 to 24 b u s In darkness
very littb 9 r m i n s
/
I
I I I I I I I
I I I I I I I
0 6 18 24
0 6 18 24
Time (Hours) 24 Hours
Figure 15. A schematic of the NO2 diurnal variation. The first 24 hours show the normal mid-latitude variation.
The situation when an air parcel crosses into polar night, when much longer periods of darkness may be
experienced. is also shown, beginning at 'last encounter with sunlight'.
(see Fig. 12), whereas by 27 January the flow had weakened somewhat and the low NO2
was most nearly confined to the polar night (Fig. 12(b)).
To test these ideas further, sensitivity calculations were carried out to examine,
firstly, the importance of the temperature dependence of the N205 photolysis in producing
the Noxon cliff, as suggested by Solomon and Garcia (1983a), and, secondly, to assess
the impact, under the conditions of this calculation, of chemistry on polar stratospheric
clouds (PSCs).
A second model run was performed to assess the sensitivity of the NOz concentration
to the temperature-dependent N2O5 absorption cross-section. The basic run, described
earlier, included the temperature-dependent N205 absorption cross-section; the second
run kept the cross-section fixed at its 273 K values. When the temperature dependence
is included, the N205 photolysis rate increases with temperature. The model temperature
in the region where NO2 peaks (= 10 mb, or 30 km) is around 200 K , and the use of the
temperature-dependent cross-section gives a slower photolysis rate than if the 273 K
cross-section were used. In consequence, the amplitude of the modelled N205 diurnal
cycle is less when temperature dependence is included and, as a result, equatorward of
the polar-night boundary the peak in the NOz concentration, which occurs just after
sunset, is lower. At 10 mb the largest difference in NO2 concentration between the two
runs was approximately 1.7 parts per billion (loy) by volume (p.p.b.v.) at close to 70"N,
accompanied by a slight decrease in CION02 and a corresponding increase in N205.
Two points are noteworthy. Firstly, it is clear that the use of the temperature-
dependent N2O5 absorption cross-section did not lead to the Noxon cliff. When the cross-
section was kept fixed at its 273 K values the shape and location of the features in the
NO2 column did not change, but the peak NO2 column just after sunset was simply
reduced by approximately 1.5 X lOI5 molecules cme2 (comparable with the calculations
of Solomon and Garcia (1983a)). Secondly, there was virtually no change in the NOz
concentrations in the polar-night region, emphasising that the reaction NO2 +
03+ NO3 + 02, followed by the formation of N205, is 'complete' in approximately 18

3-D STRATOSPHERIC CHEMISTRY MODEL 479
to 24 hours at 10 mb (Lary 1991). Thus it is the extended period of darkness experienced
by air parcels in the polar night that lead to the Noxon cliff.
Model runs were also performed to assess the sensitivity of the NOz concentration
to heterogeneous reactions on PSCs. For this simulation of January 1987 the inclusion
of reactions on PSCs made very little difference to the calculated NO2 column, simply
because the temperatures were not cold enough for their formation where the NO2
concentration peaks (= 10 mb). If the temperatures had been cold enough then sharp
spatial gradients in the NO2 column could easily have been generated by the reactions
on PSCs. This will be examined in future calculations for different periods.
Reactions on sulphate aerosols, particularly the reaction N2O5 + H 2 0 -+ 2HN03,
would decrease the NO2 concentration. However, this would be most important in the
lower stratosphere where the aerosol layer peaks, but below where NO2 peaks. In
addition, as the aerosol layer generally covers a wider area than the region of cold
temperatures producing PSCs, the modification of the NO2 column by reactions on
sulphate aerosols are themselves more likely to cover a larger area than those caused by
PSCs. Thus, on their own, reactions on sulphate aerosol seem unlikely to generate sharp
spatial gradients in NO2.
6. CONCLUSIONS
A new 3-D photochemical model of the stratosphere has been described. The model
is based on the spectral GCM being developed by the UK UGAMP. To this has been
added schemes for tracer advection, gas-phase chemistry, and photochemistry.
The model was integrated for a two-week period in January 1987. The modelled
total ozone, despite a very simple chemical initialization, agrees well with the observed
TOMS fields, demonstrating the very good dynamical performance of the model.
The model reproduces a feature resembling the Noxon cliff, with very low NO2 over
the pole. A combination of dynamical and chemical processes lead to the formation of
the cliff. Gas phase conversion of NO2 to N2OS occurs in the model, favoured by the
high ozone and high temperatures as air moves around the poleward flank of the vortex
at 10 mb. These air parcels are subject to long periods of darkness, with no photolysis
on N205, so that the formation of N205 is essentially complete in approximately 18-24
hours at 10 mb (Lary 1991).
The sensitivity to both heterogeneous chemistry and the temperature dependence
of the N 2 0 5 photolysis was studied. Under the conditions of the model run, the reactions
on polar stratospheric clouds played a negligible role. Inclusion of the N 2 0 5 temperature
dependence led to a smaller peak in the NO2 column just after sunset, with the redueed
N205 photolysis rate causing an NO2 diurnal cycle of reduced amplitude. The tem-
perature-dependent N 2 0 5 absorption cross-section did not generate the Noxon cliff in
the model, and did not significantly affect the shape or location of spatial gradients in
the NO2 column.
ACKNOWLEDGEMENTS
This work is part of the UK Universities Global Atmospheric Modelling Programme
funded by the Natural Environment Research Council. We acknowledge the support of
DGXII of the Commission of the European Communities under STEP 013 and STEP
016. Dr G. Carver is supported by UGAMP. Part of this work was carried out while Dr
Lary was supported by a Science and Engineering Research Council studentship. We
thank the ECMWF for the original version of the tracer advection code. Special thanks

480 D. J. LARY, J. A. PYLE and G . CARVER
must be given to the UGAMP core group at the University of Reading. The TOMS data
were taken from the CD ROM of gridded data for 1978-88, provided by the Goddard
Ozone Processing Team, December 1990 (P. Guimaraes and R. McPeters, Eds).
Allen, M. and Frederick, J.E.
Anderson, D. E.
Asselin, R.
Austin, J.
Austin, J . and Butchart, N .
Callis, L. B., Russell 111, J. M.,
Haggard, K. V. and
Natarajan, M.
Cariolle, D. and Deque, M.
Carver, G. C., Norton, W. and
Crutzen, P. J.
DeMore, W. B., Molina, M. J.,
Pyle, J. A.
Sander, S. P., Golden, D. M.,
Hampson, R. F.,
Kurylo, M. J., Howard, C. J .
and Ravishankra, A. R.
Dobson, G. M. B., Harrison, D. N.
and Lawrence, J .
Douglass, A. R., Jackman, C. H.
and Stolarski, R. S.
Geleyn, J. F. and Hollingsworth, A.
Geleyn, J. F., Hense, A. and
Granier, C. and Brasseur, G.
Preuss, H. J.
Grose, W. L., Nealy, J . E.,
Turner, R. E. and
Blackshear, W. T.
Hanson, D. R. and
Ravishankara, A. R.
Harwood, R. S. and Pyle, J . A.
Hoskins, B. J . and Simmons, A. J .
Johnston, H. S.
1982
1983
1982
1991
1992
1983
1986
1994
1971
1987
1990
1927
1989
1979
1982
1991
1987
1991
1975
1975
1971
REFERENCES
Effective photodissociation cross sections for molecular oxygen
and nitric oxide in the Schumann to Runge bands. J .
Afmos. Sci., 39, 2066-2075
The troposphere to stratosphere radiation field at twilight: a
spherical model. Planer. Space Sci., 31, No. 12, 1517-
1523
Frequency filter for time integrations. Mon. Weather R e v . ,
100, 487-490
On the explicit versus family solution of the fully diurnal
photochemical equations of the stratosphere. J. Geophys.
Res., 96, 12941-12974
A three-dimensional modelling study of the influence of plan-
etary wave dynamics of polar ozone photochemistry. J .
Geophys. Res., 97, 10 165-10 186
Examination of wintertime latitudinal gradients in strato-
spheric NO2 using theory and LIMS observations.
Geophys. Res. Leff . , 10, 945-948
Southern Hemisphere medium-scale waves and total ozone
disturbances in a spectral general circulation model. J.
Geophys. Res., 91, 10825-10846
Forecasting the disturbed stratospheric Arctic vortex of Jan-
uary 1992. Geophys. Res. Lett., in Press
Ozone production rates in an oxygen, hydrogen, nitrogen-
oxide atmosphere. J. Geophys. Res., 76, 731 1-7327
‘Chemical kinetics and photochemical data for use in strato-
spheric modelling’. Evaluation number 8, NASA JPL
Publication 87-41
‘Chemical kinetics and photochemical data for use in strato-
spheric modelling’. Evaluation number 9, NASA JPL
Publication W1
Measurements of the amount of ozone in the earth’s atmos-
phere and its relation to other geophysical conditions:
Part 11. Proc. R . SOC. London, A114, 521-541
Comparison of model results transporting the odd nitrogen
family with results transporting separate odd nitrogen
species. J. Geophys. R e x , 94, 9862-9872
An economical analytical method for the computation of the
interaction between scattering and line absorption of radi-
ation. Beifr. Phys. Armos., 52, 1-16
A comparison of model generated radiation fields with satellite
measurements, Beifr. Phys. Afmos., 55, 253-286
Ozone and other trace gases in the Arctic and Antarctic
regions: three-dimensional model simulations. J.
Geophys. Res., 96, 2995-3011
Modelling the transport of chemically active constituents in
the stratosphere, Pp. 229-250 in Transport processes in
fhe middle afmosphere. D. Reidel, Hingham, Mass
The reaction probabilities of CION02 and N 2 0 5 on 40% to
75% sulphuric acid solutions. J . Geophys. Res. , 96,
A two-dimensional mean circulation model for the atmosphere
below 80 km. Q. J . R . Mefeorol. SOC., 101, 723-747
A multi-layer spectral model and the semi-implicit method. Q.
J . R . Mefeorol. SOC., 101, 637-655
Reduction of stratospheric ozone by nitrogen oxide catalysts
from supersonic transport exhaust. Science, 173,517-522
17307-17314

3-D STRATOSPHERIC CHEMISTRY MODEL 48 1
Kaye, J. A. and Rood, R. B. 1989
Kuo, H. L. 1974
Lacis, A. A., Wuebbles, D. J. and 1990
Logan, J. A.
Lary, D. J. 1991
Lary, D. J. and Pyle, J. A. 1991a
1991b
Lary, D. J., Pyle, J. A., 1991
Webster, C. R. and
May, R. D.
Leovy, C. P., Sun, C-R,
Hitchman, M. H.,
Remsberg, E. E.,
Russell 111, J. M.,
Gordley, L. L., Gille, J. C. and
1985
Lyjak, L . V.
Mahlman, J. D. and Moxim, W. J.
Meier, R. R., Anderson, D. E. and
Nicolet, M.
Molina, M. J. and Rowland, F. S.
Murgatroyd, R. J.
Naujokat, B., Labitzke, K.,
Nicolet, M. and Kennes, R.
Lenschow, R., Petzoldt, K. and
Wohlfart, R. C.
Noxon, J. F.
Noxon, J. F., Whipple Jr, E. C. and
Noxon, J. F., Henderson, W. R.
Pyle, J . A.
Hyde, R. S.
and Norton, R. B.
Rood, R. B.
Rood, R. B., Kaye, J . A,,
Douglass, A. R., Allen, D. J.,
Steenrod S. and Larson, E. M.
Bumdge, D. M.
Simon, P. C., Gillotay, D.,
Simmons, A. J. and
Vanlaethem-Meurre, N. and
Wisemberg, J.
Solomon, S. and Garcia, R. R.
Thuburn, J.
1978
1982
1974
1982
1987
1986
1979
1979
1983
1980
1987
1990
1981
1988
1983a
1983b
1993
Chemistry and transport in a three-dimensional stratospheric
model: Chlorine species during a simulated stratospheric
warming. 1. Geophys. Res., 94, D1, 1057-1083
Further studies of the parametrization of the influence of
cumulus convection on large scale flow. J. Armos. Sci.,
Radiative forcing of climate by changes in the vertical dis-
tribution of ozone. J. Geophys. Res., 95, 9971-9981
‘Photochemical studies with a three-dimensional model of the
atmosphere’. PhD thesis, University of Cambridge
Diffuse radiation, twilight, and photochemistry-I. 1. Amos .
Chem., 13, 373-392
Diffuse radiation, twilight, and photochemistry-11. J. Atmos.
Chem., 13,393-406
The BLISS measurements of NOz: Some new insights.
Geophys. Res. Lea., 18, 2261-2263
31, 1232-1240
Transport of ozone in the middle stratosphere: Evidence for
planetary wave breaking. J. Atmos. Sci., 42, 230-244
Tracer simulation using a global general circulation model:
Results from mid-latitude instantaneous source exper-
iment. J. Armos. Sci., 41, 2029-2051
The radiation field in the troposphere and stratosphere from
240 to lo00 nm: general analysis. Planer. Space Sci., 30,
923-933
Stratospheric sink for chlorofluromethanes: chloxine atom
catalyzed destruction of ozone. Nature, 249, 81CL814
Recent progress in studies of the stratosphere. Q. J. R.
Mereorol. SOC., 108, 271-312
The stratospheric winter 1986/87: A major midwinter warming
35 years after they were first detected. Beilage zur Ber-
liner Wetterkarte, 56/87 SO 9/87
Aeronomic problems of the molecular oxygen photodis-
sociation. I. The Oz Herzberg continuum. Planer. Space.
Sci., 34, No. 11, 1043-1059
Stratospheric NOz, 2. Global behaviour. J. Geophys. Res., 84,
Stratospheric NOz, 1. Observational method and behaviour at
mid-latitude. J. Geophys. Res., 84, 5047-5065
Stratospheric NO2, 3. The effects of large scale horizontal
transport. J. Geophys. Res., 88, 5240-5248
A calculation of the possible depletion of ozone by chlo-
roflurocarbons using a two-dimensional model. Pure
Appl. Geophys., 118, 355-377
Numerical advection algorithms and their role in atmospheric
transport and chemical models. Reu. Geophys., 25, 71-
100
Wintertime nitric acid chemistry: Implications from three-
dimensional model calculations. J. Atmos. Sci., 47,2696-
2709
An energy and angular momentum conserving finite difference
scheme and hybrid vertical coordinates. Mon. Weather
Rev., 109, 758-766
Ultraviolet absorption cross-sections of chloro- and chloro-
fluoro-methanes at stratosphere temperatures. J. Armos.
Chem., 7 , 107-135
On the distribution of nitrogen dioxide in the high latitude
stratosphere. J. Geophys. Res., 88, 5229-5239
Simulation of NO, partitioning along isobaric parcel traject-
ories. J. Geophys. Res., 88, 5497-5501
Use of a flux-limited scheme for vertical advection in a GCM.
Q. J. R. Meteorol. SOC., 119, 469-487
5067-5076

D. J. LARY, J. A. PYLE and G. CARVER
Tiedtke, M. 1983
Tibaldi, S., Palmer, T. N.,
Vaughan, G. and Price, J. D.
Webster, C. R., May, R. D.,
1990
Brankovic, C. and Cubasch, U.
1991
1990
Tourni, R. and Pyle, J . A.
WMO 1986
‘The sensitivity of the time-mean large scale flow to cumulus
convection in the ECMWF model’. Pp. 297-316 in Pro-
ceedings of the ECMWF workshop on convection in
large-scale models, Reading
Extended range predictions with ECMWF models. 11: Influ-
ence of horizontal resolution on systematic errors and
forecast skill. Q. J . R. Mereorol. Soc., 116, 835-866
On the relation between total ozone and meteorology. Q. J.
R. Mereorol. SOC., 117, 1281-1298
Active nitrogen partitioning and the nighttime formation of
N205 in the stratosphere: Simultaneous in situ measure-
ments of NO, NOz. HNO,, 0, and N20 using the BLISS
diode laser spectrometer. J . Geophys. Res., 95, 13851-
13866
Atmospheric ozone 1985. Assessment of our understanding
of the processes controlling its present distribution and
change. World Meteorological Organization Global
Ozone Research and Monitoring Project, Report No. 16