Journal of Geophysical Research: Atmospheres

Evaluation of the chemical transport model Oslo CTM2 with focus on arctic winter ozone depletion

Authors


Abstract

[1] The global chemical transport model Oslo CTM2, driven by European Centre for Medium-Range Weather Forecasts international Integrated Forecast System meteorology and including tropospheric and stratospheric chemistry, is described and evaluated against measurements by satellite-based instruments, ozone sondes, and aircraft. A new heterogeneous chemistry and microphysics scheme capable of reproducing severe ozone depletion events is described. The Oslo CTM2 simulates ozone measurements well, overestimating by less than 20% in the lowermost stratosphere, while underestimating slightly at higher altitudes. In the upper troposphere, the spatial and temporal variation is well simulated, although a 30–50% overestimation is seen. In order to demonstrate ozone depletion in the Arctic by heterogeneous processes, the ozone loss during the cold 2004/2005 Arctic winter stratosphere is compared to the warm 2000/2001 Arctic winter. Enhanced ozone loss is shown to have occurred at eight Arctic stations. From January to March the accumulated ozone loss amounts to up to 1 ppmv, with a maximum ozone loss rate of 32 ppb/d. Ozone sonde measurements are well reproduced for both years.

1. Introduction

[2] Atmospheric scientists have been using computer models for more than 20 years to help understand the physics and chemistry of the atmosphere. Our ability to reproduce the observed atmosphere should reflect our understanding of the atmosphere. In order to do so, all the key physical and chemical processes must be included in the model. The main limitation on this work has always been computing power, and even though the recent years have provided faster and better computers, atmospheric models can still easily be too complex for practical research if injudicious choices are made as to what level of complexity physical processes are modeled and the choice of horizontal and vertical resolution. Until recently, the solution to this has usually been to focus on only the region of interest, leaving out processes or using highly simplified schemes, e.g., using only tropospheric chemistry and no stratospheric chemistry for studying the troposphere, or leaving out the troposphere when studying the stratosphere. Today, more and more models (CCMs and CTMs) are being improved to include comprehensive chemistry and physics of both the troposphere and the stratosphere [e.g., Rotman et al., 2004; Chipperfield, 2006; Kinnison et al., 2007], as has been done for the Oslo CTM2.

[3] The access to real meteorological data based on numerical weather prediction models makes CTMs powerful tools for studying atmospheric chemistry, allowing more detailed comparisons with measurements to be carried out. It is important to recognize that the quality of the meteorological reanalysis is important for the reproducibility of measurements. However, the increasing amount of observational data available for the UTLS (e.g., satellite, aircraft and sonde data) gives an excellent opportunity to validate the performance of global-scale models, in particular how the models represent physical and chemical processes occurring in the UTLS.

[4] Parameterization of physical processes occurring on smaller scales than the model resolution can also be tested out more realistically. By including key atmospheric processes, CTMs are well suited for process studies and for comparisons with measurements when using real meteorology. Representing UTLS processes realistically is crucial for understanding the distribution of species in this region, and in this study we focus on ozone, which is one of the most important species in the UTLS and the middle atmosphere. The important processes include stratospheric to tropospheric exchange (STE), which requires accurate transport, and removal due to activation of free radical reactants which is linked to polar stratospheric clouds (PSCs) formed at high latitudes [Meilinger et al., 2001]. Chemical reactions on particles such as PSCs and background aerosols (heterogeneous chemistry) temporarily remove NOx and activate chlorine. When reaching sufficient sizes the particles slowly sediment due to gravity, causing both denitrification and dehydration. These physical processes determine the depth and extent of any ozone hole after the end of a polar winter. Since nitrogen has the potential to deactivate chlorine [Fahey et al., 2001; Tolbert and Toon, 2001], denitrification will enhance chlorine activation and thus ozone depletion.

[5] To study the upper troposphere and lower stratosphere (UTLS), one needs to consider both tropospheric and stratospheric chemistry as well as transport processes. The Oslo CTM2 is a global CTM including comprehensive chemistry for both the troposphere and stratosphere. Recently, our parameterization for particle microphysics of PSCs and heterogeneous chemistry has been updated. Parameterizations for the different types of PSCs, a particle size distribution and particle sedimentation are now included in the Oslo CTM2, and it is therefore well suited to study processes occurring in the UTLS region.

[6] In this study we validate this new scheme and evaluate the model performance by comparing simulations with measurements for the year 2000, and then investigate the UTLS region during the Arctic winter 2004/2005, the coldest Arctic stratosphere on record.

2. Oslo CTM2

[7] The Oslo CTM2 is an off-line model driven by meteorological data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecast System (IFS) model (http://www.ecmwf.int/research/ifsdocs/). The meteorological data are given on a 3 h basis, produced for each day by a 36 h forecast with 12 h of spin-up, initialized from the analysis at noon (1200 UTC) the previous day (discussed previously by, e.g., Wild et al. [2003]). Using forecasts rather than analyses gives a more dynamically self-consistent data set, and has been shown to give more realistic transport [e.g., Stohl et al., 2004; Scheele et al., 2005]. Also the use of 3 hourly meteorological data instead of, e.g., 6 h has been found to improve the transport [e.g., Bregman et al., 2006]. In the IFS forecasts a spectral resolution of T319 is applied. The horizontal resolution of the Oslo CTM2 can be varied between T21 (5.5 × 5.5°), T42 (2.8 × 2.8°), T63 (1.9 × 1.9°) and 1 × 1°, into which the IFS spectral fields are truncated. The IFS data available as gridded data are averaged into the model grid. Sigma pressure hybrid coordinates are used in the vertical, extending in 40 layers from the surface up to 2 hPa (the uppermost layer mass center is at 10 hPa). In the tropopause region the vertical resolution varies between about 0.8 km at high latitudes and about 1.2 km at low latitudes, and above 100 hPa the resolution is 20 hPa. Advective transport is calculated using the highly accurate and low diffusive second-order moments scheme [Prather, 1986]. The parameterization of deep convection is based on the Tiedtke mass flux scheme [Tiedtke, 1989], while boundary layer mixing is treated according to the Holtslag K profile scheme [Holtslag et al., 1990]. The numerical integration is carried out using operator splitting, sequentially simulating every process during 1 h.

[8] Dry deposition of species follows Wesely [1989], while wet deposition and washout are calculated on the basis of the ECMWF data for convective activity, cloud fraction, rain fall and on the solubility of the species [Berglen et al., 2004]. Both large-scale and convective washout processes are represented.

[9] To make the Oslo CTM2 suited for studying processes in the UTLS, the original tropospheric model [Sundet, 1997; Berntsen and Isaksen, 1997] was extended to include comprehensive chemistry for the stratosphere as well [Gauss, 2003]. A heterogeneous chemistry scheme [Carslaw et al., 1995] and the Fast-J2 method for calculation of photo dissociation coefficients [Wild et al., 2000; Bian and Prather, 2002] was included, and the vertical resolution was improved. The parameterizations of lightning and aircraft emissions, both important for the nitrogen budget in the UTLS, were refined.

[10] The Oslo CTM2 has now been improved with a new scheme for microphysics and heterogeneous chemistry, to better represent the formation of PSCs, including denitrification and dehydration. All the chemistry modules will be discussed in detail in the next two sections.

[11] The Oslo CTM2 has previously been applied in model/model comparisons and tested against observations [Grini et al., 2002; Brunner et al., 2003; Isaksen, 2003; Eskes et al., 2003; Gauss et al., 2003b; Berglen et al., 2004; Isaksen et al., 2005; Brunner et al., 2005; Gauss et al., 2006].

2.1. Emissions

[12] The emission data sets for source gases were prepared in the European Union (EU) project POET (Precursors of Ozone and their Effects in the Troposphere) [Granier et al., 2005; Olivier et al., 2003]. These emissions are available on 1 × 1° resolution for the years 1990–2000. In this study we use the data for 2000 for the simulations of 2004/2005 as well as for 2000.

[13] The lightning source is based on data from Price et al. [1997], with the original global output value scaled down to 5 Tg(N)/a (where a is years). To conserve the meridional distribution and the fraction between marine and continental areas in the Oslo CTM2, these data are interpolated into the horizontal model grid and summed up zonally and separately for marine and continental areas for each month. Then the monthly and zonally integrated data are redistributed according to the lightning frequencies over marine and continental areas by the use of the ECMWF convection data. For each longitude, latitude and meteorological time step we define a “lightning activity,” which is a measure of the lightning frequency, calculated according to the formulas given by Price et al. [1997], relating lightning frequency to cloud top heights.

[14] The cloud top height is represented as the height of the convection cell. This procedure has previously been described by Berntsen and Isaksen [1999]. For continental and tropical marine areas, lightning emissions are distributed vertically within a column according to Pickering et al. [1998], who provided the fraction of lightning emissions within a column in 1 km vertical resolution. If such an interval contains two or more Oslo CTM2 layers, the fraction is subdivided according to layer thickness. Another approach is used for marine areas poleward of 30°, where we distribute the total lightning source according to the mass fraction in the grid box with respect to the tropospheric column mass.

[15] Aircraft emissions are taken from the TRADEOFF 2000 revised standard scenario (as described briefly by Gauss et al. [2006]), and are interpolated online into the Oslo CTM2 grid at each time step. Chemical conversions may take place in the plume behind the aircraft, and cannot be resolved by a coarse model resolution. We use look-up tables created by the NILU aircraft plume model [Kraabøl et al., 1999] to allow the inclusion of plume-scale processes. The conversion into other nitrogen species (NO, NO2, NO3, HNO3, N2O5, HO2NO2, HNO2, and PAN) is calculated every hour as a function of time, latitude, background concentrations of NOx and ozone, temperature and turbulent conditions. During this partitioning the total mass of the emissions is conserved. In addition, the ozone change in the plume as a function of the initially emitted NOx molecules is estimated. Plume chemistry is found to reduce maximum perturbations in zonal mean ozone by about 20% [Gauss et al., 2003a].

2.2. Chemical Scheme

[16] The evolution of chemical compounds at any point in the model domain is controlled by transport, emissions, deposition and chemical production and loss,

equation image

where A is the change due to transport, E is emissions, D is deposition, Pc is chemical production, and Lc is chemical loss. Transport is important through the whole model domain. However, this does not apply for all chemical conversions; some species are of negligible chemical importance in the troposphere or in the stratosphere. This makes it possible to combine two chemical schemes, one for the troposphere and one for the stratosphere; species with negligible chemical conversions in the stratosphere are not treated in the stratospheric chemistry scheme. Species that are not reactive in the troposphere are similarly not treated in the tropospheric scheme. Only compounds important for both tropospheric and stratospheric chemistry are treated in both schemes. To select which chemical scheme to apply, we use the NCEP reanalysis tropopause pressure retrieved every 6 h. The NCEP tropopause data is provided by the NOAA-CIRES Climate Diagnostics Center (CDC), Boulder, Colorado, USA, from their web site (http://www.cdc.noaa.gov/) on a 2.5 × 2.5° horizontal grid, and is interpolated into the horizontal Gaussian grid of the Oslo CTM2. For each column in the horizontal grid, the top of the tropospheric domain is defined as the uppermost grid box with center pressure greater than the NCEP tropopause pressure. The closest grid box above is the lowermost grid box of the stratospheric domain. The tropospheric and the stratospheric chemistry schemes are applied in the tropospheric and stratospheric domains, respectively. Recently, the NCEP tropopause has been tested against a tropopause on the basis of potential vorticity (PV), revealing small differences except in the Southern Hemisphere ozone hole conditions. Using the PV-based tropopause, the ozone column is lowered about 3% in the Antarctic polar vortex, which is due to a 2–3 km lower tropopause level. We conclude that for this Arctic study the choice of tropopause does not affect the simulations to a large extent, although we recognize that for future studies the PV-based definition should be used.

[17] The tropospheric chemistry scheme is run with a numerical time step of 15 min (5 min for OH/HO2/RO2 reactions), contains 51 species and takes into account 86 thermal reactions, 17 photolytic reactions and 2 heterogeneous reactions (which are independent of the new heterogeneous chemistry). It includes detailed hydrocarbon chemistry and has been thoroughly tested [Berntsen and Isaksen, 1997; Kraabøl et al., 1999; Berglen et al., 2004; Brunner et al., 2003].

[18] The stratospheric chemistry scheme is an extension of the scheme used by Stordal et al. [1985] for the Oslo 2-D model, and was later updated to include heterogeneous chemistry [Isaksen et al., 1990] before it was included in the 3-D Oslo SCTM-1 [Rummukainen et al., 1999] and the Oslo CTM2. 55 species and 7 families are included, and a total of 159 reactions (104 thermal, 47, photolytic and 8 heterogeneous), which are integrated with a numerical time step of 5 min. Of theses species, 17 are also treated in the tropospheric scheme. The heterogeneous chemistry scheme is a part of the stratospheric chemistry, and is described in detail along with the microphysics in section 3.

[19] All species in the Oslo CTM2 (amounting to 97 including families) are listed in Appendix A, stating which components are transported (77 of 97 species) and whether they are of chemical importance to tropospheric chemistry, stratospheric chemistry or both (i.e., in which chemical scheme they are treated). A list of chemical reactions treated by the Oslo CTM2 can be found in Appendix B. Reaction rate constants in this study are taken from Jet Propulsion Laboratory [2003], unless otherwise is described in the text.

[20] For transported species of tropospheric interest only, e-folding times are defined in the stratosphere in order to mimic stratospheric removal. The e-folding time is chosen to match available observations in the stratosphere. For compounds of importance for stratospheric chemistry, the upper boundary conditions are fixed by climatological mixing ratios provided as geographical zonal means by the Oslo 2-D model. Ideally, the 2-D data should be mapped into the 3-D model by the use of equivalent latitude, since the polar vortex may be dislocated from the pole. Sensitivity studies have previously been carried out, indicating that this will have a small effect on the model simulations. The transported species also need lower boundary conditions: Emissions of gases with a long tropospheric lifetime, e.g., CFCs, HCFCs and N2O, are approximated by keeping mixing ratios in the lower troposphere fixed on the basis of World Meteorological Organization's [2003] suggestions for the year to be simulated. Short-lived species of negligible tropospheric importance, treated in the stratosphere only, are defined in the lowermost 2 km from multiyear simulations of the Oslo 2-D model, approximating surface emissions suggested by the WMO. Hence, the cross-tropopause gradients of species treated only in the stratosphere are set up by the interactions between transport and the stratospheric chemistry. It is important to note that the chemistry below the upper boundary and above the lower boundary/surface is running freely.

[21] Water vapor is treated differently in the tropospheric and stratospheric chemistry schemes. Tropospheric water vapor is derived from ECMWF data, while stratospheric water vapor is calculated assuming that the sum of hydrogen (as H2) in water vapor, methane and hydrogen is constant (H2O = Hsum − 2 × CH4 − H2). Hsum = 6.97 ppmv is chosen for present-day simulations, while a larger value is chosen in future simulations, reflecting increases in methane. The stratospheric water vapor may be reduced by the formation of PSC2 at low temperatures (see section 3). In studies of aircraft impact, H2O emitted by aircraft in the stratosphere is added to the water vapor.

[22] In both schemes the numerical integration of chemical kinetics is done applying the Quasi Steady State Approximation (QSSA) [Hesstvedt et al., 1978], using three different integration methods depending on the chemical lifetime of the species.

3. Microphysics and Heterogeneous Chemistry

[23] An important issue of atmospheric chemistry modeling is heterogeneous chemistry (chemical reactions on particles), and therefore also the microphysics of particles. Different types of such particles exist, and treated in the Oslo CTM2 are “traditional” aerosols, which are taken from satellite data (referred to as background aerosols), and also PSCs. The old microphysics and heterogeneous scheme did not take PSC2 and sedimentation into account, resulting in too low chlorine activation. A more sophisticated scheme has now been implemented [Galin et al., 2007; de Zafra and Smyshlyaev, 2001; S. P. Smyshlyaev et al., Chemistry-climate modeling of gas and particle transformation in polar regions, submitted to Izvestiya Atmospheric and Oceanic Physics, 2007, hereinafter referred to as Smyshlyaev et al., submitted manuscript, 2007], representing formation and evolution of PSCs, including denitrification and dehydration through sedimentation. Since theses processes are important for the activation of chlorine and thus the ozone depletion, the new parameterization will allow for studies where heterogeneous chemistry is important.

3.1. Formation and Evolution of PSCs

[24] A lognormal size distribution of particles is assumed, and their formation is simulated through detailed calculation of particle densities, depending on temperature, pressure and partial pressures of the relevant species. The variables involved in the formation comprise gaseous HNO3, solid HNO3, H2O, and total H2SO4, as well as air pressure and temperature. H2SO4 is not included as a chemical species in the model, and is therefore calculated from the background aerosol data following Hanson et al. [1994], where the sulfuric acid mass fraction in a sulfuric acid solution is calculated assuming a plane surface in equilibrium with the ambient water vapor partial pressure. From this mass fraction the volume density and the mass mixing ratio of H2SO4 are derived, assuming a particle median radius of 0.0725μm with a standard deviation of 1.86.

[25] We distinguish between PSC1 (both nitric acid trihydrate (NAT or PSC1a) and supercooled ternary/binary solutions (PSC1b)–which consist of liquid H2SO4/HNO3/H2O) and PSC2 (frozen H2O possibly coated by NAT). Volume and HNO3/H2SO4 partitioning are calculated according to Carslaw et al. [1995] as explained by Smyshlyaev et al. (submitted manuscript, 2007). On the basis of partial pressures of H2O, HNO3 and H2SO4, the composition of the ternary solution is calculated by applying Henry's law constants. If H2SO4 in the solution is diminishingly small, the composition is calculated as a binary HNO3/H2O solution according to Hanson and Mauersberger [1988]. The solution is then treated as PSC1a below a certain temperature TMIX,

equation image

which is the freezing temperature for a mixture of HNO3 and sulfuric acid tetrahydrate, as described by Fox et al. [1995]. The partial pressures of the subscribed species are given in Torr. Once frozen, the particle is assumed to be solid until the melting temperature TMELT,

equation image

given by Zhang et al. [1993] is reached (TMELT > TMIX). The volume of frozen aerosols is subject to an equilibrium approach, where the volume once formed is assumed constant until melting temperature is reached. The formed aerosols are then treated as PSC1 below 205 K, and as supercooled background aerosols from 205 to 215 K. According to Carslaw et al. [1995] the aerosol is assumed to be a binary solution of H2O and H2SO4 above 215 K.

[26] PSC2 are allowed to exist below 190 K, and their H2O content at the given temperature is found by calculation of the saturated vapor pressure of H2O according to Marti and Mauersberger [1993]

equation image

where the pressure is given in Pascal. An ice density of 928 kg m−3 is assumed, from which the PSC2 volume per unit volume of air is calculated. PSC2 may possibly be coated with frozen HNO3. Supersaturated HNO3 is calculated according to Hanson and Mauersberger [1988] assuming a supersaturation corresponding to a supercooling of 3 K, and freezes as PSC2 coating at temperatures lower than TICE. It is important to note that below TICE, the already formed frozen aerosols (NAT) will remain as NAT and still be subject to sedimentation, leaving PSC1 and PSC2 in coexistence. Frozen particles are allowed to exist up to TMELT, but in that way they are assumed to be PSC2 below 190 K, NAT-coated ice from 190–205 K and frozen sulfate aerosols from 205 K to TMELT.

[27] The temperature of particles does not change as fast as their surroundings [e.g., Voigt et al., 2000]. Existing particles coming from a very cold area may still be cold enough to persist in solid state even though the surrounding temperature has become too high for formation. For particles entering warmer air we therefore introduce the temperature history by a 5 d running mean temperature, and particles do not melt until the running mean is greater than the melting temperature. However, when a particle enters even colder air, the particle is assumed to adjust its temperature more quickly, growing according to the actual temperature (not the running mean).

3.2. Lognormal Particle Size Distribution

[28] Different lognormal size distributions are assumed for the three different types of particles. The distributions consist of 40 different particle sizes ranging from 0.014 to 19 μm in particle radius, and are calculated as

equation image

where ri is the radius of the each particle bin, and r0 is the mean radius of the distribution with a standard deviation σ. The mean radii of PSC1a, PSC1b and PSC2 are r0 = 0.4, 0.07, and 5.0 μm, respectively, and the corresponding standard deviations are σ = 1.6, 1.9, and 1.8. The surface area densities (SAD) are calculated according to the size distributions.

3.3. Sedimentation

[29] The sedimentation velocity of a particle is taken from look-up tables as a function of particle size and density [Kasten, 1968]. The velocities depend on the particle sizes and determine how far the particles will fall, redistributing their masses accordingly. Because of the different size distributions, the particles are sedimented separately and for each of the respective particle size bins. The corresponding amounts of liquid/solid HNO3 and H2O are sedimented accordingly, causing irreversible denitrification and dehydration. Particles sedimented into warmer air will be subject to melting and evaporation if the temperature is high enough.

[30] Although the formation of particles and the heterogeneous chemical reactions on them are constrained to the stratosphere, PSCs and existing solid/liquid HNO3 and H2O may be sedimented into the troposphere, where they are allowed to evaporate when the melting temperature is reached. In the troposphere gaseous HNO3 may further be removed by washout processes.

3.4. PSC and Aerosol Chemistry

[31] The calculation of surface area densities of PSC1, PSC2 and aerosols is the essential part of the new microphysics scheme, since the reaction rate coefficients for reactions on particles depend on the surface area densities. Here, PSC1a and PSC1b are treated the same way below 205 K (subscript PSC1), and as aerosols above 205 K (subscript AER). For PSCs the rate coefficient is given as

equation image

where b is the reaction name and γ1,b and γ2,b are uptake coefficients for PSC1 and PSC2, respectively, given in Table 1 [Jet Propulsion Laboratory, 2003]. APSC1 and APSC2 are the calculated surface area densities of PSC1 and PSC2, respectively, and Wa is the average molecular velocity of the nitrogen or chlorine/bromine component, a, reacting on the particle. Reactions on PSCs will convert NOx to solid HNO3, which stays on the particle. However, if the air then becomes subsaturated, HNO3 will evaporate until saturation.

Table 1. Reactions on PSC1, PSC2, and Aerosolsa
NameReactions on PSCsγ1,bγ2,bγ3,ba
  • a

    The subscript 1, 2, and 3 represent PSC1, PSC2, and aerosols, respectively. For reactions on PSCs, (s) denotes that the products stay on the particle (solid state). Otherwise the products are in gaseous form. PSC1b is treated as PSC1a below 205 K and as aerosols above. The γ3,bd, γ3,bc and, γ3,ec values are calculated and are described in the text.

ADN2O5 + H2O → 2HNO3 (s)0.00040.0250.1N2O5
ACN2O5 + HCl → ClONO + HNO3 (s)0.0030.030.0N2O5
BDClONO2 + H2O → HOCl + HNO3 (s)0.0040.03γ3,bdClONO2
BCClONO2 + HCl → Cl2 + HNO3 (s)0.230.26γ3,bcClONO2
ECHOCl + HCl → Cl2 + H2O (s)0.140.26γ3,ecHOCl
FDBrONO2 + H2O → HOBr + HNO3 (s)0.0060.260.4BrONO2
FCBrONO2 + HCl → BrCl + HNO3 (s)0.30.50.8BrONO2
GCHOBr + HCl → BrCl + H2O (s)0.250.10.2HOBr

[32] Given a temperature between 205 and 215 K, the amount of aerosols (PSC1b) is used for calculation of the surface area density of background aerosols (cold aerosols). This may modify the already existing background data, so at present, the aerosol surface area is reset to the original background data each time step, to keep it consistent with calculating H2SO4 from the background data.

[33] The chemical reactions on background aerosols are responsible for heterogeneous conversion of NOx to gaseous HNO3, and are listed in Table 1. Their reaction rate coefficients are given according to the equation

equation image

where AAER is the surface area density, and the reactive uptake coefficient γ3,b is taken from [Jet Propulsion Laboratory, 2003], except for γ3,fd [Hanson and Ravishankara, 1995], γ3,gc [Hanson and Ravishankara, 1994], and γ3,fc [Hanson et al., 1996]. For background aerosol area densities less than 10−10 cm−1γ3,bd, γ3,bc and γ3,ec are equal to 0.01, 0.008, and 0.001, respectively. Otherwise a background aerosol particle radius of a = 10−5cm is assumed and γ3,bd and γ3,bc are given by Hanson and Ravishankara [1994], while γ3,ec is given by Hanson et al. [1994] (assuming a H2SO4 wt% minimum of 30% and that the radius is small compared to the reactodiffusive length).

3.5. Heterogeneous Chemistry Scheme and Ozone Depletion

[34] Daily zonal mean of the simulated total ozone column for the year 2000 is presented in Figure 1, given in Dobson units (DU). In lower latitudes, the total ozone column value is between 220 and 250 DU, while the largest total ozone columns are modeled during spring both in high northern and high southern latitudes. The ozone hole at the South Pole is well simulated, beginning to form in July/August, reaching its lowest ozone values in October (139 DU), and then diminishing in November as the spring maximum takes over. Although this is a little higher than the observed minimum, it should be noted that the data are daily and zonally averaged, and that locally lower values may be encountered. In fact, the lowest modeled value on an hourly basis is 130 DU on 16 September.

Figure 1.

Zonal daily mean Dobson values from the Oslo CTM2, T42L40 resolution, year 2005 (with a minimum of 138.8 DU and a maximum of 487.5 DU).

[35] The use of data from the Oslo 2-D model in the uppermost model layer introduces some uncertainties in the ozone column. The 2-D model is not consistent with the Oslo CTM2 meteorology and heterogeneous chemistry, and may therefore be too high, especially for ozone hole conditions. Some ozone is also present above the Oslo CTM2 upper boundary, suggested to be about 5–15 DU by the Oslo 2-D model. The monthly averaged 2-D data also introduces the slight discontinuities in Figure 1, where the ozone in the uppermost layer changes from month to month.

[36] A sensitivity study excluding PSCs reveals that the effect of PSCs is a 30% reduction of the daily averaged Dobson value during Southern Hemisphere ozone hole conditions (Figure 2). At the South Pole, this reduction corresponds to an ozone reduction of about 70% at 18 km altitude (Figure 3) on an hourly basis, which is a little low compared to measurements [e.g., World Meteorological Organization, 2003]. Although the model resolution is T42, it may be coarse enough to cause too much mixing across the polar vortex, thus reducing the effect of PSCs. Also, the treatment of advection at the highest latitudes may affect this and should be investigated.

Figure 2.

Modeled ozone column reduction due to PSCs for the year 2005.

Figure 3.

The reduction of ozone due to PSCs at the South Pole in the austral spring 2005.

[37] As the austral spring begins and the ozone hole starts to form, ClO increases because of the reaction of active chlorine with ozone. The maximum modeled daytime ClO value for this simulation was 1330 pptv on the 10 September. Also BrO increases somewhat during this period, with a daytime maximum of about 12 pptv in September–October (not shown).

[38] Gaseous NOy consists mostly of HNO3, and most HNO3 will eventually condense during PSC formation. Solid HNO3 will also form chemically on PSCs from other nitrogen components (see section 3.4 and Appendix B). As a result, gaseous NOy is reduced to almost zero and irreversible denitrification is caused by sedimentation.

[39] Water vapor depletion during PSC2 formation gives a gaseous H2O monthly zonal mean minimum value of 1.8 ppmv in August in the Southern Hemisphere. The daily zonal mean at the South Pole shows a stratospheric minimum of 1.4 ppmv (not shown), reflecting effective dehydration in the stratosphere.

4. Model Evaluation

[40] We evaluate the Oslo CTM2 against measurements from satellites, ozone sondes and aircraft. The simulations are completed using T42 horizontal resolution for the year 2000.

4.1. Comparison With Satellite Measurements of Ozone

[41] In this study we compare SAGE II profiles [Wang et al., 2002] with the model profiles from the corresponding model time and location. Averaging the profiles in a spatial and temporal area of, e.g., 15° latitude during a month, provides comprehensive comparison and also allows for statistical considerations. For all locations in the spatial and temporal domain, averages of each model layer thickness are calculated, and then the model and satellite data are averaged into this spacing. Comparisons of several latitude intervals and time periods with good satellite coverage are shown in Figures 4 and 5, where the mean modeled (blue) and measured (red) values are plotted as vertical lines, representing the vertical extent of each of the explained layers. The standard deviations (σ) for these layers are indicated as horizontal lines. The satellite mean is plotted on the SAGE II vertical resolution (500 m) as the black line, with corresponding standard deviations given in orange. Given to the right is the percentage difference between model and measurement averages for each model level. Our focus is on the UTLS, so only levels above 8 km are shown. Since the uppermost model level is set by the boundary conditions on a monthly basis, the uppermost level is left out.

Figure 4.

Vertical profiles from SAGE II versus the Oslo CTM2. Monthly means of 15° latitude intervals for year 2000, showing SAGE II mean, SAGE II on CTM2 mean height, CTM2 on CTM2 mean height and standard deviations (see text for description) for (a) February at 30–45°N, (b) February at 15–30°N, (c) March at 75–60°S, (d) March at 45–60°N, (e) April at 75–90°N, and (f) April at 60–75°N.

Figure 5.

Same as Figure 4 but vertical profiles from SAGE II versus the Oslo CTM2 for (a) May at 45–30°S, (b) June at 30–15°S, (c) June at 45–60°N, (d) July at 60–75°N, (e) November at 30–45°N, and (f) December at 75–60°S.

[42] Typically 50 to 200 profiles are averaged, and in general there is good agreement between Oslo CTM2 and the measurements, also when the standard deviations are considered. The Oslo CTM2 mostly reproduces measured values well within 1σ. It should be noted that SAGE II data are less reliable below 15 km, with up to 50% underestimation of ozone below the tropopause [Wang et al., 2002]. Nonetheless, we plot the data from 8 km upward. The Oslo CTM2 tends to overestimate in the upper troposphere in low latitudes (Figures 4b and Figure 5a), and in the Southern Hemisphere higher latitudes where also the lower stratosphere is overestimated (Figure 5f). The latter could be due to a too fast ozone hole recovery, or because of a too shallow ozone hole, which again could be due to too high upper boundary conditions in this region (not shown). In the Northern Hemisphere the simulations look good (Figures 4a, 4d4f, and 5c5e), with maximum differences smaller than 20% above 12 km.

[43] The upper boundary conditions are noticed to be too low at several temporal and spatial locations outside the polar vortices, and it should be noted that this may lead to more UV radiation reaching lower altitudes, possibly resulting in slightly increased ozone values further down. Sensitivity studies to test this suggest that this effect is small.

[44] As already noted, the measured and modeled means lie mostly within each other's standard deviations. In the latitude interval 45°S–15°N, standard deviations of the satellite data are generally smaller compared to other latitude intervals, indicating less variability in that area. Otherwise, the model variability is usually about the same as for the satellite data.

[45] Although the model profiles are taken at the same spatial and temporal location as the measured profiles, in some cases the model resolution may cause a model profile and a measurement profile to cover different synoptic meteorological conditions. This could affect the means and standard deviations. However, for each model layer, the mean values of the model and the satellite are mostly positively correlated from one time interval to the next, indicating that the model captures most of the meteorological features which could affect ozone. Also T21 horizontal resolution reproduces the SAGE II measurements in much the same way (not shown), although with slightly larger standard deviations, which could be somewhat due to the model not capturing the correct synoptic conditions.

[46] In general, the total ozone columns are well simulated, and as an example of a short-term feature mainly forced by the meteorological data, we show the models ability to reproduce the ozone minihole over Europe appearing in late November 2000 [Semane et al., 2002]. The simulated total ozone column on the 28 November 2000 (1030 local time) is shown in Figure 6a, along with the column measured by the GOME instrument [Valks et al., 2003] (Figure 6b). The percent difference between Oslo CTM2 and GOME is shown as white contours in Figure 6b. Specifically, the local maximum over the Mediterranean Sea is well modeled, and the ability to reproduce the minihole features confirms the accuracy of the meteorological data. Still, the model tends to underestimate the maximum values in the Northern Hemisphere higher latitudes, and overestimate in the Southern Hemisphere. Partly this may be explained by the Oslo 2-D boundary data, since the 2-D model comprises a different heterogeneous chemistry scheme. Over Siberia the Oslo CTM2 overestimates up to 44%, which is probably also due to the coarse vertical resolution. An increase in the vertical resolution will give more insight in this. The Southern Hemisphere overestimation may be due to the too shallow ozone hole modeled, followed by a too fast ozone recovery in the austral spring. However, total ozone column values are improved substantially compared to T21 resolution and compared to the model version with the previous heterogeneous chemistry scheme (not shown), especially at the South Pole. Preliminary tests with increased vertical resolution also show improvements on this, and will be investigated further.

Figure 6.

Total ozone from (a) Oslo CTM2 and (b) GOME T42 resolution on the 28 November 2000. White contours on Figure 6b show percent difference between Oslo CTM2 and GOME.

4.2. Comparison With Ozone Sonde Vertical Profiles

[47] During the EU Third European Stratospheric Experiment on Ozone (THESEO 2000) campaign (http://www.nilu.no/projects/theseo2000/) ozone vertical profiles were measured using ozone sondes at different locations in the Northern Hemisphere, mostly in Europe. Model profiles at the time and model grid points of each sonde are provided, and for each observation location the profiles are averaged over a month in the same way as for the satellite profile averages (Figure 7). Again the Oslo CTM2 does a good job reproducing the ozone values and the variability–standard deviations are overlapping quite well, although the model boundary conditions (top layer) are underestimated.

Figure 7.

Monthly averaged vertical profiles from the THESEO campaign year 2000, compared with the Oslo CTM2, T42L40 resolution for (a) Prague in January, (b) Ny Ålesund in February, (c) Lerwick in February, and (d) Payerne in December.

[48] In Figure 7b we see that the model tends to overestimate ozone in the 18–24 km region, although this effect is much smaller for T42 than for T21 resolution (not shown). Increasing the vertical resolution might give a better insight in this overestimation. As for the satellite comparison, the variation of the uppermost model layer is introduced through the averaging process, but the variation is smaller here because the location is fixed, and is why the whole uppermost layer is not shown. These comparisons also suggest that the upper boundary conditions given by the Oslo 2-D model are too low.

4.3. Comparison With Measurement of Ozone and Water Vapor by Airbus In-Service Aircraft Measurements

[49] The Measurement of Ozone and Water Vapor by Airbus In-Service Aircraft (MOZAIC) aircraft measurements [Marenco et al., 1998] are made by instruments onboard conventional aircraft. In order to evaluate the Oslo CTM2 against MOZAIC measurements, linear interpolations (in space and time) to the MOZAIC measurement locations were done, allowing flight-by-flight comparisons. The MOZAIC measurements are mainly confined to below 12 km, but cover both the troposphere and the lowermost stratosphere, depending on latitude and season. We compare the Oslo CTM2 with these measurements flight by flight, which is a rigorous test for a model, both in terms of model resolution and of transport, chemistry and physics.

[50] A spatial resolution of T42 in 40 vertical layers and a temporal resolution of 1 h corresponding to the model operator splitting time step is rather coarse compared to the dense spacing of MOZAIC measurements during a flight. Consequently, while the main structure is well reproduced by the model, the variability of the MOZAIC measurements on very short timescales cannot be resolved. Because of this, we average the flight data confined by intervals of 500 m height or by a time period short enough for the aircraft to cover less than one model grid box (chosen to be 15 min), visualized in Figure 8.

Figure 8.

MOZAIC versus Oslo CTM2 for four different flights. Blue solid line is Oslo CTM2, blue dashed line is Oslo CTM2 in T21 resolution, red dashed line is MOZAIC measurements, black solid line is the CTM2 tropopause (NCEP), and the black dashed line is aircraft altitude at a given time.

[51] Flight measurements are well reproduced, although the model does not always capture the horizontal or vertical gradients when crossing the tropopause several times on a relatively short distance. This is evident in Figure 8c, where the model overestimates in the upper troposphere and underestimates in the stratosphere. T42 horizontal resolution should be able to reproduce gradients across ∼16° latitude, unless the vertical gradients are not captured correctly. The slacker gradients in Figure 8c may be due to a too coarse model vertical resolution, but could also be due to too weak subsidence of stratospheric air in the meteorological data. Also seen in Figure 8 are model results using T21 resolution, clearly showing the benefit of increasing the resolution to resolve these gradients. A further increase of the model resolution is expected to resolve the gradients better, and this will be investigated in the future. However, when the measured gradients cover larger distances the model works well.

[52] For each flight the correlation between the modeled and measured data was calculated, along with mean values and standard deviations. Each correlation can be plotted against the standard deviation in a Taylor diagram [Taylor, 2001]. Focusing on the UTLS, and selecting data points above 8 km altitude, we have calculated the correlations between modeled and measured flights for February, March, June, and December 2000. To be able to plot all correlations in the same plot (Figure 9), the standard deviation for each modeled flight has been normalized to the standard deviation of the corresponding measured flight. Some points lie outside the dashed circular line, meaning that these modeled flights have larger standard deviations than the measurements, whereas most points lie on the inside (smaller standard deviations than the measurements). Larger standard deviations are in general found in December, when the model tends to overestimate ozone in the lower stratosphere as well as the upper troposphere (e.g., Figure 8d). Lower standard deviations are in general found in summer (June). This is expected because of the model resolution, and corresponds well with what we have seen in Figure 9 where the sharp gradients are underestimated. A possible reason is also that winter conditions are better described by the meteorology than are summer conditions, and this will be a subject for future studies. Correlations (the angular placement of the points) show to what extent the model capture variations along the flights. For flights with few data points and the shorter flights staying below 8–9 km, this comparison is troublesome either because of few data points or also because of low ozone values combined with relatively large variations. For longer flights with good data coverage the correlations in general lie between 0.7 and 0.9, confirming the ability of the Oslo CTM2 to simulate ozone aircraft measurements in the UTLS on a global scale. The discussions of the mean values are omitted here, since the data are normalized. However, the skill score suggested by Taylor [2001] is plotted, showing that on the whole, the model shows good skill, especially in winter conditions.

Figure 9.

Taylor diagram of flight-by-flight comparisons between Oslo CTM2 and MOZAIC measurements. Flights for February (black), March (blue), June (orange), and December (red).

5. Arctic Winter 2004/2005

[53] Arctic winter ozone loss has been subject for several studies [e.g., Chipperfield et al., 1993; Bojkov et al., 1998; Schulz et al., 2000; Rex et al., 2002], and cold temperatures have been found to be crucial to ozone loss [e.g., World Meteorological Organization, 2003]. So far, the coldest Arctic stratosphere on record has been during the winter 2004/2005, a time of frequently observed PSCs. The ozone loss of this winter has therefore been studied and quantified to a large extent [Streibel et al., 2005; Manney et al., 2006; Rex et al., 2006; Jin et al., 2006; Feng et al., 2007; Grooß and Müller, 2007; D. R. Jackson and Y. J. Orsolini, personal communication, 2007]. We apply the Oslo CTM2 to simulate a cold stratosphere (2005) and a warm stratosphere (2001), to test the strength of the model in simulating heterogeneous chemistry and ozone loss in the Arctic.

[54] For January–March several vertical model profiles have been compared with ozone sondes, confirming how well the Oslo CTM2 reproduces the observations (Figure 10). The Oslo CTM2 profiles are plotted as vertical blue bars extending through each model layer, while the red bars represent the Oslo CTM2 output without PSCs. The simulations started from the same initial conditions at the 1 January. Similarly, a number of vertical profiles for the warm Arctic stratosphere during winter 2001 (January–March) is shown in Figure 11, where the Oslo CTM2 again shows its modeling capability. Especially interesting is the reproduction of the height levels with lower ozone values, e.g., Figures 10k and 11e11j. Since both the simulations with and without PSCs reproduce these features, they are not a result of heterogeneous chemistry, but most probably because of transport processes. As for the previous evaluations, because of artifacts from the upper boundary conditions, we cut the vertical axis at 25 km altitude.

Figure 10.

A random selection of ozone sonde measurements and the corresponding Oslo CTM2 modeled vertical profiles of Arctic stations in 2005, for January, February and March. AL, Alert; EU, Eureka; NA, Ny Ålesund; RS, Resolute; SC, Scoresbysund; SO, Sodankyla.

Figure 11.

A random selection of ozone sonde measurements and the corresponding Oslo CTM2 modeled vertical profiles of Arctic stations in 2001, for January, February, and March. Same abbreviations as used in Figure 10.

[55] Using the same initial field for the 3 month simulations with and without PSCs, and the fact that heterogeneous chemistry is more effective in March because of more sunlight, are two of the reasons why the two simulations show greater differences in March than in January. Also, the ozone at one station is dependent on heterogeneous chemistry at other locations through transport. Because of the Northern Hemisphere planetary Rossby waves, chlorine and bromine activated in dark areas may be transported into sunlit areas where they may deplete ozone, and then the ozone depleted air can be transported back into an area without sunlight. This is more likely to happen in late winter since the sun is reaching farther north, illuminating larger parts of the air coming from cold areas.

[56] In Figures 12a, 12b, 12c, and 12d we see the effect of PSCs (PSC minus noPSC) on ozone, BrO, ClO, and the modeled chemical loss at Ny Ålesund during March, respectively, given by Oslo CTM2 hourly data. For this month the maximum ozone difference between the simulation with and without PSCs (effect of PSCs) is 352 ppbv. Chlorine is the main component responsible for polar ozone loss in the lower stratosphere during spring, producing ClO. ClO is only formed when sunlight is present (when Cl2 is photolysed to 2Cl), reaching maximum values in March when the sun has fully returned. ClO reacts with itself, reactivating chlorine [Molina and Molina, 1987] which again can destroy ozone. Similar reactions exist for bromine; however, BrO also reacts with ClO, a reaction much faster than the ClO-ClO reaction, so that BrO may be quickly reduced. The BrO-ClO reaction reactivates Cl and Br (either directly or by forming BrCl, which is photolysed by sunlight). When PSCs are present, the amount of ClO increases and thereby the loss of BrO is increased. In this process the BrO loss may be enhanced more than the formation of BrO is enhanced, resulting in a BrO reduction due to PSCs (negative values in Figure 12b). Recently, Pope et al. [2007] suggest that the rate of ClOOCl photo dissociation is lower than the Jet Propulsion Laboratory value used in most models, and that some missing reaction chain is needed to explain the chlorine activation seen during ozone hole formation. This will be a subject for future studies.

Figure 12.

The effect of PSCs on (a) ozone, (b) BrO, (c) ClO, and (d) on the chemical loss of ozone at the location, given by the Oslo CTM2 on an hourly basis at Ny Ålesund, for March 2005. Contours on Figure 12d are the surface area density of PSCs (0.1 and 1μm2/cm3).

[57] At Ny Ålesund, ClO starts to form at 20–24 km in February and reaches maximum values in March at about 18 km. The correspondence of the changes in ClO and ozone loss rate is striking (Figure 12). For the chemical loss of ozone from hour to hour we see a maximum ozone loss due to PSCs of about 2 ppbv/h in March. Accumulated this amounts to about 20 ppbv/d, which is comparable with other studies [Chipperfield et al., 1993; Streibel et al., 2005]. Although values up to about 60 ppbv/d have been reported for the cold Arctic stratosphere during the 1999/2000 winter [Rex et al., 2002], Manney et al. [2006] reported that the ozone loss was not as large in 2004/2005, and only slightly larger than produced by the Oslo CTM2. The small negative values of the ozone loss change due to PSCs, between 21 and 24 km in Figure 12d, are due to a slight decrease in NO2 due to heterogeneous chemistry. PSCs (and cold aerosols) increase the conversion of N2O5 to HNO3, and thereby reducing NO2 and its impact on ozone loss at that altitude. It may also be that more NO2 is being bound as ClONO2 at the end of the day because of more ClO, which coincides well with the time of day for the ozone loss reduction.

[58] At some hours ozone is somewhat reduced even though the chemical loss is very small or even absent, indicating incoming transport of ozone depleted air, an effect which can be seen at other Arctic locations as long as PSCs are not present at the location.

[59] When looking at the altitude range between 12 and 21 km at eight Arctic stations in 2005 and 2001, the daily maximum changes of ClO due to PSCs are shown in Figure 13, along with the maximum daily ozone losses due to PSCs (for the height level of largest loss). For the cold 2005 stratosphere (Figures 13a and 13b), the maximum daily ozone loss is about 3–9 ppbv/d for January, increasing until mid-March (up to 30–33 ppbv/d for Andøya and Sodankyla). For the warm 2001 stratosphere (Figures 13c and 13d), however, there is almost no effect of PSCs. There are small changes in ClO, and the maximum daily ozone loss mainly varies between 0 and 7 ppbv/d, except for a short period in February with cold temperatures above northern Scandinavia, which were cold enough for PSC formation (and enhanced ClO) at Andøya, Sodankyla and to some extent at Scoresbysund.

Figure 13.

Daily maximum change in ClO due to PSCs and the maximum daily accumulated ozone loss for eight Arctic stations, as modeled by the Oslo CTM2 for (a and b) 2005 and (c and d) 2001.

[60] Listed in Table 2 are the eight Arctic stations and their accumulated chemical ozone losses for each of the uppermost nine model layers, which span from about 11–25 km (grid center). The losses are accumulated from January to March (JFM) 2005, and the loss values are given in percent of the mean ozone mixing ratios in the respective layers (mean for January–March). The total chemical ozone loss ranges from about 2% to up 39% at the level of max depletion (18 km), which is due to both heterogeneous processes on PSCs and loss due to NOx. The latter is most prominent at 22–24 km after vortex break-up. For Andøya the loss due to NOx starts to increase at 12 March and dominates after 19 March. When omitting the days after 15 March, the total loss amounts to 33% instead of 39%. The maximum total ozone loss at these stations corresponds to about 1 ppmv between 12 and 21 km, which is somewhat more conservative than some studies [Rex et al., 2002; Jin et al., 2006; Feng et al., 2007], but correspond well to others [Grooß and Müller, 2007; D. R. Jackson and Y. J. Orsolini, personal communication, 2007].

Table 2. Accumulated (January, February, and March) Stratospheric Ozone Loss at Different Altitudes for Eight Arctic Stations, Given as Percent of the Mean Ozone Value at the Given Locationa
HeightNy ÅlesundAndøyaAlertSodankylaEurekaResoluteScoresbysundThule
  • a

    Each layer altitude is the mean of the layer grid center altitudes for JFM at the location. From top to bottom the table gives total ozone loss in JFM 2005, ozone loss due to PSCs JFM 2005, total ozone loss JFM 2001, and ozone loss due to PSCs JFM 2001.

Total Chemical Ozone Loss January–March 2005 (Percent of Mean Ozone)
24.4 km−11.5−16.8−8.7−18.4−10.0−15.0−14.6−12.0
20.3 km−10.7−28.5−5.3−27.7−6.6−12.5−22.5−10.9
18.0 km−14.8−38.5−6.4−38.8−7.7−13.0−25.0−12.0
16.3 km−12.1−27.2−6.2−28.9−8.3−14.2−17.9−11.7
14.9 km−9.1−23.1−5.5−23.2−7.9−14.0−14.4−10.5
13.8 km−6.6−19.1−4.7−18.0−7.1−12.6−11.8−9.2
12.8 km−5.0−13.5−4.1−12.9−6.1−11.3−9.3−8.0
11.9 km−3.6−7.3−3.6−7.6−5.2−10.0−6.2−6.6
11.1 km−2.6−3.7−3.2−4.6−4.5−8.4−3.1−5.8
 
Chemical Ozone Loss Due to PSCs January–March 2005(Percent of Mean Ozone)
24.4 km−1.1−3.8−0.2−4.3−0.8−2.8−3.8−2.0
20.3 km−6.3−17.8−1.9−17.2−2.1−4.4−13.4−5.1
18.0 km−9.8−26.1−3.1−26.7−3.2−5.0−15.1−5.8
16.3 km−7.4−16.6−2.9−18.4−3.6−6.2−9.0−5.6
14.9 km−4.9−13.8−2.3−14.0−3.6−6.5−6.6−4.8
13.8 km−2.6−10.9−1.6−9.8−2.9−5.8−4.7−3.8
12.8 km−1.2−6.6−0.9−6.0−1.8−4.5−3.2−2.7
11.9 km−0.4−2.5−0.4−2.6−1.0−3.2−1.9−1.4
11.1 km−0.1−1.1−0.2−1.3−0.4−2.0−0.9−0.7
 
Total Chemical Ozone Loss January–March 2001 (Percent of Mean Ozone)
25.2 km−8.2−15.4−6.6−15.5−7.1−10.0−13.9−10.1
20.9 km−4.9−14.9−3.4−15.9−4.6−7.6−11.7−6.1
18.4 km−5.5−15.1−3.5−15.6−4.6−7.8−11.3−6.3
16.6 km−4.7−13.5−3.2−14.0−4.1−7.1−9.5−5.7
15.2 km−4.3−11.9−3.0−12.2−3.9−6.5−8.3−5.6
14.0 km−4.2−10.7−2.9−10.8−3.8−6.3−7.5−5.5
13.0 km−4.4−10.1−3.0−9.8−3.8−6.4−7.2−5.5
12.0 km−4.8−9.2−3.3−8.9−4.0−6.3−7.7−5.3
11.2 km−4.9−8.3−3.5−8.1−4.2−6.3−8.2−5.2
 
Chemical Ozone Loss Due to PSCs January–March 2001 (Percent of Mean Ozone)
25.2 km−0.1−1.6−0.0−1.6−0.0−0.0−0.8−0.1
20.9 km−0.2−4.2−0.0−4.9−0.1−0.3−2.2−0.2
18.4 km−0.5−4.4−0.1−5.0−0.2−0.8−2.1−0.5
16.6 km−0.6−3.8−0.1−4.4−0.2−0.7−1.5−0.6
15.2 km−0.6−3.10.1−3.4−0.1−0.6−0.8−0.8
14.0 km−0.5−2.40.2−2.50.1−0.5−0.4−1.0
13.0 km−0.2−2.00.1−1.80.1−0.5−0.2−0.8
12.0 km−0.1−1.30.1−1.10.1−0.4−0.2−0.3
11.2 km−0.0−0.70.1−0.60.1−0.2−0.1−0.1

[61] The difference in ozone losses between the simulations with and without PSCs for JFM 2005 are given in Table 2b, and show that PSCs are responsible for ozone losses up to about 26% at, e.g., Andøya and Sodankyla. As already noted, the increase in loss due to PSCs may be accompanied by a slight reduction of NO2 and thus the ozone loss due to NO2.

[62] At the other stations, PSCs contribute between 0 and about 15% ozone loss. The non-PSC ozone loss is mainly due to NO2 when the sun returns, although bromine chemistry and heterogeneous chemistry on aerosols are also somewhat responsible. Heterogeneous processes and bromine chemistry both increase with lower temperature, increasing ozone loss at lower temperatures. For the warm Arctic stratosphere of 2001, the total ozone losses (Table 2c) are smaller than for 2005, and in contrast to the cold Arctic stratosphere, the ozone losses due to PSCs for 2001 (Table 2d) are mainly in the range of 0–1%, except from the stations experiencing temperatures cold enough for PSC formation, with ozone loss up to 5% due to PSCs.

6. Conclusions and Future Directions

[63] We have presented the Oslo CTM2, a global chemistry transport model including detailed physical processes and comprehensive chemistry for the troposphere and stratosphere, as well as heterogeneous chemistry on PSCs and aerosols, and the microphysics of PSCs. The model is capable of reproducing severe ozone depletion events, such as the ozone hole and miniholes, and compares well with measurements in the UTLS region.

[64] When compared to MOZAIC measurements, the Oslo CTM2 has a slight tendency to underestimate ozone in the lower stratosphere and to overestimate it in the upper troposphere. Where sharp gradients are encountered, the model underestimates up to 50% due to the model resolution or too weak subsidence of stratospheric air. Increasing the resolution will give further insight in this. The overestimation in the upper troposphere and underestimation in the lower stratosphere is to some degree confirmed by ozone sonde measurements, and may be due to the coarse vertical resolution. Preliminary studies with 60 layer IFS meteorological data show promising improvements due to increased vertical resolution and the coverage of the whole stratosphere.

[65] The profile comparisons suggest that the model upper boundary values of ozone most probably are too low; however, the Oslo CTM2 does a good job reproducing most of the profiles. Also vertical profiles measured by SAGE II are very well reproduced by the Oslo CTM2. Increasing the horizontal resolution to T42 is shown to improve the simulations, giving the model data a finer structure and a better fit to observations. A further increase is expected to give further improvements.

[66] The Oslo CTM2 is also capable of reproducing measurements of both cold and warm Arctic stratospheres, giving chemical ozone losses due to PSCs comparable with other studies. PSC formation is dependent on temperatures and the amount of HNO3, H2O and H2SO4, although a cold temperature is the most crucial requirement. The ECMWF data does not only provide these temperatures, but also a realistic circulation including the polar vortex and the planetary waves, also allowing intrusion of southern air and the transport of activated chlorine into sunlit areas. The combination of realistic meteorology and the comprehensive chemistry of the Oslo CTM2, results in the ability to reproduce well chemical events originating from meteorological processes.

[67] A crucial point in the PSC calculations seems to be the requirement of the sulfuric acid amount–which is not included as a species in the model. In the current formulation of the model, the sulfuric acid is calculated on the basis of satellite measurements of background aerosols. Another possible solution is to implement a sulfur scheme for the stratosphere. At the moment only a tropospheric sulfur scheme is available, which is not used in these simulations.

[68] In addition to increasing the vertical resolution, the upper boundary for the Oslo CTM2 will be updated in order to address the general underestimation at the uppermost model layer. The amount of measurements of different species has increased dramatically in the recent years, and will be extensively used in future evaluations. As meteorological data up to near present are now available, more detailed evaluation and studies of interannual variations will be completed.

Appendix A:: Chemical Species in the Oslo CTM2

[69] All components included in the Oslo CTM2 are listed in Table A1, with their CTM2 number (Nr), whether they are transported (T = Y) or not (T = N) along with the chemical domain of interest (C): T is troposphere only, S is stratosphere only and B is both. Families are shown in italic, and their molecular weights are not listed, since they are only used for conversion to and from mass space for transport (i.e. they can be arbitrary as long as the same value is used both ways). The families not transported are only treated in chemistry.

Table A1. Chemical Species of the Oslo CTM2
NameT/CRemarks
O3Y/B 
NOXY/T= NO + NO2 + NO3 + 2N2O5 + HO2NO2 + PAN
NOZN/B= NO3 + N2O5 (only treated in chemistry)
HNO3Y/B 
PANxY/T= PAN + CH3COO2
COY/B 
C2H4Y/T 
C2H6Y/T 
C3H6Y/T 
C4H10Y/T 
C6H14Y/T 
C6HXRY/Tm-xylene
CH2OY/B 
CH3CHOY/T 
H2O2Y/B 
CH3O2HY/B 
HO2NO2Y/B 
CH3COYY/TCH3COCOCH3
CH3COXY/TCH3COC2H5
IsopreneY/TC5H8
HO2Y/B 
CH3O2N/B 
C2H5O2N/T 
C4H9O2N/T 
C6H13O2N/T 
CH2O2OHN/T 
CH3COBN/TCH3COCH(O2)CH3
CH3XXN/TCH3CH(O2)CH2OH
AR1N/Tfirst RO2 radical from the reaction of m-xylene with OH
AR2N/Ta C-5 carbonyl compound formed from the reactionof AR1 with NO
AR3N/Ta C-5 RO2 radical formed from the reaction of AR2 with OH
ISOR1N/Tfirst RO2 radical from the reaction of isoprene with OH
ISOKN/Tmethylvinylketone (MVK) + methacrolein (MACR)
ISOR2N/TRO2 radical formed from MVK + OH or MACR + OH
HCOHCOY/T 
RCOHCOY/T 
CH3XY/TCH3COO2
O(3P)N/B 
O(1D)N/B 
OHN/B 
NO3Y/B 
N2O5Y/B 
NOY/B 
NO2Y/B 
O3NON/TO3 minus NO (only treated in chemistry)
CH4Y/B 
DMSN/Tdimethyl sulphide, (CH3)2S
C3H8Y/T 
C3H7O2Y/T 
AcetoneY/TCH3C(O)CH3
CH3CODY/TCH3COCH2(O2)
MCFY/SCH3CCl3
HCFC-22Y/SCF2HCl
CFC-11Y/SCFCl3
CFC-12Y/SCF2Cl2
CCl4Y/S 
CH3ClY/S 
N2OY/S 
ClxY/S= Cl + ClO + OHCl + ClONO2 + 2Cl2 + OClO + BrCl + ClOO + 2Cl2O2
NOx_strY/S= NO + NO2 + NO3 + 2N2O5 + ClONO2 + BrONO2 + HO2NO2
SOY/S= O3 + O(1D) + O(3P)−NO−Cl−Br
HClY/S 
ClyY/S= Clx + HCl + HCls
H2N/S 
H2ON/S 
SHY/S= H + OH + HO2 + 2H2O2
CH3BrY/S 
H-1211Y/SCF2ClBr
H-1301Y/SCF3Br
BryY/S= Br + BrO + BrONO2 + OHBr + HBr + 2Br2 + BrCl
H-2402N/SC2F4Br2
CFC-113Y/SCCl2FCClF2
CFC-114Y/SCClF2CClF2
CFC-115Y/SCClF2CF3
HNO3sY/Ssolid phase (heterogeneous chemistry)
H2OsN/Ssolid phase (heterogeneous chemistry)
HCFC-123Y/SCF3CHCl2
HCFC-141Y/SCFCl2CH3
HCFC-142Y/SCF2ClCH3
HN/S 
ClN/S 
ClON/S 
OHClY/S 
ClONO2Y/S 
Cl2Y/S 
OClOY/S 
BrN/S 
BrON/S 
HBrY/S 
BrONO2Y/S 
OHBrY/S 
Br2Y/S 
ClOOY/S 
Cl2O2Y/S 
BrClY/S 
NOy_strY/S= NOx_str + HNO3

Appendix B:: Reactions in the Oslo CTM2

[70] CTM2 takes into account all reactions listed below (i.e. both tropospheric and stratospheric chemistry). Reactions in bold type are included in both the tropospheric and the stratospheric modules. Reaction rate constants in the Oslo CTM2 are continuously updated according to the Jet Propulsion Laboratory, and is used in the Jet Propulsion Laboratory's [2003] study. The chemistry schemes are described sections 2.2. and 3. The “(s)” indicates species staying on the particle, and applies only for reactions on PSCs/STS.

B1. Photolysis Reactions

B1.1. Tropospheric Chemistry Module
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B1.2. Stratospheric Chemistry Module
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image

B2. Chemical Reactions in the Tropospheric Module

equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image

B3. Chemical Reactions in the Stratospheric Module

B3.1. Ox Reactions
equation image
equation image
B3.2. O(1D) Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.3. HOx Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.4. NOx Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.5. Hydrocarbon Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.6. ClOx Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.7. BrOx Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image
B3.8. Instantaneous Reactions
equation image
equation image
equation image
B3.9. Thermal Decomposition Reactions
equation image
equation image
equation image
equation image
B3.10. Heterogeneous Reactions
equation image
equation image
equation image
equation image
equation image
equation image
equation image
equation image

Acknowledgments

[71] The authors acknowledge the EU MOZAIC project (EVK2-1999-00141) for data collection and analysis at CNRS, Forschungszentrum Jülich, University of Cambridge, and Météo-France. We also acknowledge MOZAIC support from EADS Airbus and the airlines Air France, Deutsche Lufthansa, Sabena, and Austrian Airlines who carry the MOZAIC instruments free of charge and perform maintenance. The work of S. Smyshlyaev was supported by the Russian Foundation for Basic Research, Project 06-05-64821. His work included developing and testing of the new heterogeneous chemistry scheme during the INTAS project (collaboration between University of Oslo and the Russian State Hydro-meteorological University).