SPARC logo (45 Ko)

Stratospheric Processes And their Role in Climate
A project of the World Climate Research Programme

Home Initiatives Organisation Publications Meetings Acronyms and Abbreviations Useful Links


Report on the Arctic Ozone Loss Workshop
Potsdam, Germany, 4-6 March 2002

Markus Rex, Alfred Wegener Institute, Potsdam, Germany (
Darin Toohey, University of Colorado, Boulder, USA (
Neil R.P. Harris, EORCU, Cambridge, UK (

With contributions from the rapporteurs: A. Dörnbrack, S. Godin, G. Millard, M.L. Santee, H. Schlager, A. Schulz, B.-M. Sinnhuber, R. Stimpfle, R. Ruhnke, and Y. Terao


Over the past decade tremendous progress has been made toward quantifying and understanding ozone loss in the Arctic stratosphere. Today a variety of approaches exist to quantify the degree of chemical loss over the course of an Arctic winter. Some have been used in a consistent way for many years and have produced a record of the interannual variability. On the other hand, a wide range of chemical models have been used to understand the processes in the wintertime Arctic stratosphere and to calculate the degree of ozone loss.

An active scientific discussion has started about the level of maturity of up to date chemical models of the polar stratospheric chemistry. How well are observations of the ozone loss rate reproduced? How complete is our current quantitative understanding of the involved chemical and microphysical processes? Are discrepancies between model results and observations larger than the combined known uncertainties?

In the published literature the answers to these questions are controversial. Over the next few years one of the major challenges for the stratospheric research community will be to predict the future of the Arctic ozone layer in a scenario of decreasing stratospheric halogen loading and possible changes in climate. A solid assessment of our ability to reproduce currently observed ozone losses with model calculations is indispensable to determine the requirements for future research and to correctly interpret the reliability of model based predictions.

To address these issues, the Arctic Ozone Loss Workshop was held on 4-6 March in Potsdam, Germany, hosted by the Alfred-Wegener-Institute for Polar and Marine Research. The conveners (the authors of this summary) were joined in a programme committee by G. Amanatidis, M. Kurylo , P. Newman, and J. Pyle. About 70 scientists from Europe, the US, Japan, Russia and New Zealand have participated. The workshop was mainly based on poster presentations and discussion sessions. It was part of DYCHO, a research project within the German AFO-2000 program and of QUOBI, an EC funded research project, which are both part of the EU research cluster SOLO. The workshop was supported by SPARC.In his opening remarks M. Kurylo highlighted the role of the workshop as an arena for focused discussions where the current status and the requirements for future research should be identified. The workshop covered five sessions that are summarized in the following.

Ozone Loss: Observations

The opening session was on the empirical determination of ozone loss. A variety of techniques have been developed in recent years:

a) those which use explicit transport calculations based on meteorological data,
b) those which use long-lived tracers.

There are now estimates of ozone loss based on measurements for nearly every winter since 1988/89. Scientists involved with almost all these analyses were present at the workshop. F. Goutail presented multi-annual results from comparing a passive ozone tracer in the chemical transport model (CTM) REPROBUS with ozone observations from the SAOZ ground-based network and from the Polar Ozone and Aerosol Measurement (POAM) satellite instrument. The same technique was used by G. Hansen with the SLIMCAT CTM and ozone lidar measurements. G. Manney showed ozone loss estimates for various years based on measurements from the Microwave Limb Sounder (MLS) and ozone advection by trajectory calculations. M. Rex reported one decade of results from Match campaigns, a lagrangian approach to observe chemical ozone loss. In presentations based on POAM data [K. Hoppel], ozonsonde profiles [G. Braathen and M. Rex] and airborne lidar measurements [W. Grant] chemical ozone losses were derived using the vortex average technique. V. Dorokhov used ozone soundings from Siberia to estimate ozone loss. S. Hayashida presented results from a trajectory mapping approach based on measurements from the Improved Limb Atmospheric Spectrometer (ILAS). A. Schulz derived the temperature dependency of ozone loss rates from Match data. Y. Terao applied the Match approach to ILAS data, and S. Tilmes presented a homogenised multi-annual re-evaluation of ozone/tracer correlation studies based on data from the HALogen Occultation Experiment (HALOE).The discussion centred on the level of agreement between these techniques. The general result is illustrated by the four coloured boxes in Figure 1, which indicate different degrees of reliability of current observationally based techniques.

Figure 1: Schematic illustrating the level of reliability of current approaches to estimate chemical ozone losses from ozone observations.
(For a better resolution of the images, please click on the plot or contact the SPARC Office)

In region (i) many techniques have been used with consistent results. All agree that in cold Arctic winters the main ozone loss occurs in this region. Recent work comparing results for well-defined time periods and geographical locations indicate quantitative agreement between the various independent approaches better than 20% (95% confidence). The largest local ozone loss that was observed in the Arctic so far occurred in winter 1999/2000. In Figure 2 this loss is compared with typical Antarctic conditions.

Figure 2: Illustration of the degree of ozone loss in the Arctic winter 1999/2000, the coldest Arctic winter on record, compared with the typical ozone loss over Antarctica.
(a) Concentration profile of O3 during late winter from ozonesonde observations between 20 and 30 March 2000 (red). The profile of O3* (blue), i. e. the abundance of O3 expected in the absence of any chemical loss, is estimated by allowing the early winter vortex average O3 mixing ratio profile from ozonesondes to descend by amounts based on cooling rates from the SLIMCAT CTM. The difference between the profile of O3* and O3 (hashed) is an estimate for the accumulated chemical loss of ozone during the winter, based on the vortex average approach. The black lines represent a typical late March ozone profile above and below the vertical region where significant ozone loss occurred. Adapted from Rex et al., Chemical depletion of Arctic ozone in winter 1999/2000, J. Geophys. Res., in press.
(b) Illustration of the typical degree of ozone loss in the Antarctic. Typical mid-winter (15 July 1997, blue) and late winter (13 October 1997, red) ozone profiles measured inside the Antarctic polar vortex at the Neumayer station at 71°S. This panel is for illustration only and no attempt has been made to correct for transport effects, which are weaker in the Antarctic compared to the Arctic.
(For a better resolution of the images, please click on the plot or contact the SPARC Office)

In region (ii), fewer studies have been carried out, with some disagreement as to whether significant ozone loss is occurring. However, the majority of analyses showed clear evidence that it does indeed occur, predominantly at the higher levels of region (ii), i.e. above the flight levels of the ER-2. Losses up to 500 ppbv (at around q=500 K) were observed in cold winters with daily loss rates of up to 40 ppbv/day.

There are considerably fewer studies for the regions marked (iii) and (iv). Uncertainties associated with heating rates are greater in these regions and rigorous horizontal mixing particularly in region (iv) is an insuperable challenge for all currently known approaches.


Ozone changes in the Arctic winter and spring are the result of both chemistry and transport. Presentations in this session focused on the relative contributions of both to the variability and long-term changes of the total ozone column in the Arctic [S. Anderson, R. Kivi, M. Proffitt, M. Rex, M. Weber], long-term changes and the accuracy of meteorological parameters [G. Braathen, B. Knudsen, respectively], the influence of mixing and net transport on ozone abundances [A. Dörnbrack, R.-S. Gao, G. Manney, Y. Orsolini], the processes that govern the initial ozone distribution in the early polar vortex [R. Kawa], and on the processes that determine the temperature in the Arctic stratosphere [P. Newman].

In her overview, G. Manney stressed the importance of interannual variability of stratospheric dynamics on the abundance of ozone in the Arctic polar vortex. She discussed the effects of wave activity, diabatic descent and tropopause heights on Arctic ozone fields based on meteorological data and MLS observations.

P. Newman showed that March temperatures in the polar vortex are correlated with wave fluxes into the stratosphere during winter. Models show this qualitatively, but not quantitatively, raising concerns about the ability to accurately simulate polar temperatures and, hence, chemical ozone losses in a changing climate.

B. Knudsen showed that the variability in Arctic total column ozone is correlated with the chemical ozone loss within the vortex. M. Rex quantified the dynamical supply of ozone to the Arctic and the wintertime chemical loss for the past decade. Both concluded that dynamics and transport contribute about equally to the observed interannual variability of the Arctic ozone columns. B. Knudsen also reported that about 75% of the Arctic ozone decline during the nineties was due to chemistry and discussed the accuracies of temperature fields from various assimilation models.

The discussion for this session focused on the uncertainties in descent, their impact on ozone loss studies, and what steps can be taken to improve model simulations of polar dynamics.

The general view of the participants was that it could be shown that chemistry accounts for approximately half of the interannual variability of column ozone and that the largest fraction of the decrease of ozone since the eighties was due to chemical depletion. It was noted that large discrepancies remain in temperature analyses, and it will be difficult to achieve the accuracy desired for simulating highly non-linear chemical processes.

Halogen Observations

Abundances of ClO and BrO and their temporal variability are critical in order to calculate chemical ozone destruction. This session featured presentations that addressed issues related to the accuracies of halogen measurements [L. Avallone, M. Santee, and R. Stimpfle], the partitioning of inorganic chlorine [K. Kreher, J.-P. Pommereau, H. Oelhaf, M. Santee, I. Wohltmann], ozone loss rates based on measured ClO and BrO [K. Pfeilsticker, and F. Stroh] and long-term trends in BrO [D. Toohey].

During SOLVE/THESEO-2000 there were a number of opportunities when in situ and remote measurements of ClO overlapped. L. Avallone showed that reported abundances of ClO typically agreed within the combined uncertainties when the measurements were nearly co-located and differences in solar zenith angles are taken into account. But in individual cases differences of ~30% were found, which would imply a difference in chemical ozone loss of over 50%.

Abundances of ClO measured from the NASA ER-2 during SOLVE/THESEO-2000 based on preliminary data were smaller than those measured by other techniques. R. Stimpfle reported that, based on post-mission laboratory calibrations, the final ClO data will increase by 10-25% (depending on temperature). He also showed that this change reduces the apparent inorganic chlorine budget discrepancy that had been discerned from the preliminary data. M. Santee showed that the better vertical resolution in version 5 of MLS algorithms results in a significant reduction in the abundances retrieved at 20 km altitude.

The agreement between measurements of BrO by different techniques was addressed during the discussion session. Remote techniques report abundances of BrO in the polar vortex that are ~10-20% larger than in situ measurements. K. Pfeilsticker showed that remote observations of BrO in 2000 are consistent with a total bromine budget of 23+/-2.5 ppt, whereas previous in situ observations have supported a budget between 15 and 18 ppt. D. Toohey showed that BrO abundances have increased in the Arctic vortex at a rate of ~3% per year since 1989, thus explaining a significant fraction of the discrepancies between the remote and in situ observations (the latter having been carried out between 1992 and 1995). Because abundances of chlorine source gases are expected to decrease faster than those of the bromine source gases, the relative role of bromine in polar ozone destruction will further grow in importance.

K. Kreher showed that column abundances of ClO, HCl, and ClNO3 over Scott Base, Antarctica, are reproduced well by the SLIMCAT CTM during the period of major ozone destruction. This indicates that the budget and partitioning of inorganic chlorine are reasonably well understood during the period of maximum activation, but the model does underestimate ozone loss both in the Antarctic [ K. Kreher] and in cold Arctic years [I. Wohltmann]. Using MLS data for 7 years, M. Santee showed that ClO abundances are more variable in the Arctic than in the Antarctic, but that during the cold mid-1990s years in the Arctic, chlorine activation was nearly as extensive as over Antarctica. F. Stroh presented results from the TRIPLE balloon flights in 2000 that can be used to infer photochemical rate parameters. The results were broadly consistent with model results based on the current recommendations of the JPL panel.

K. Pfeilsticker reported that some simultaneous observations of OClO and NO2 suggest high abundances of active chlorine could coexist with non-negligible abundances of NOx, in contrast to expectations based on photochemical models. Such an observation could imply activation of chlorine in darkness, and the possibility for unrecognised reactions that destroy ozone.

W. Bloss presented results from a recent laboratory study indicating that the rate constant for the reaction ClO+ClO is ~15-30% larger than currently recommended by the JPL panel at low temperatures (< 200 K) that prevail during the period of chlorine activation and ozone loss.
The general conclusion from the halogen session was that the various techniques used to detect ClO and BrO provide comparable results, but that specific intercomparisons would be very useful for reducing uncertainties that limit the ability to bring more consistency to the observations of ozone loss and calculations of chemical ozone destruction. The possible coexistence of NOx and active chlorine and the implications for ozone destruction need further investigation, especially in the context of unexplained ozone losses that have been observed at high solar zenith angles in mid-winter.

Polar Stratospheric Clouds and Denitrification

The primary factor that determines the rate of chlorine deactivation in springtime and, hence, the cessation of ozone loss is the abundance of reservoirs of reactive nitrogen (NOy) that can be depleted by sedimenting PSC particles – a process termed denitrification. Presentations in this session covered topics ranging from observations of denitrification [R. Bevilacqua, A. Kleinböhl, M. Santee], observations of particles [D. Fahey, M. Müller], particle microphysics [ K. Drdla, B. Luo], and the relationship between denitrification and ozone loss [K. Drdla, R.-S. Gao, R. Ruhnke].

Measurements obtained during SOLVE/THESEO-2000, summarised by D. Fahey, demonstrated that denitrification can occur by formation of a small number of very large (> 10 mm diameter) NAT particles. However, the mechanism by which these particles form is unclear. B. Luo showed that initial particle growth and sedimentation in the supersaturated lower edge region of NAT ‘mother clouds’ downstream of ice clouds could be the source of such particles, but argued that the accuracies in temperature fields necessary for accurate model descriptions of this process may be difficult to achieve.

R. Bevilacqua presented a multi-year climatology of polar stratospheric clouds based on POAM data. By analysing the threshold temperature for PSC formation some information about the degree of denitrification can be inferred, indicating that denitrification may be more common in the Arctic than previously thought.

K. Drdla argued that 50% of the ozone loss observed in the Arctic in 2000 resulted from heterogeneous reactions that occurred after March 1, but that chlorine reactivation is very effective on liquid aerosols, not on NAT particles.

In the break-out discussion the importance of future studies of the PSC formation process, the evolution of the particle composition and phase during their life cycle, and the extent of denitrification they are causing were stressed. These studies will benefit from a much higher level of coordination between different observational efforts than in the past.

Ozone Loss (2): Models

The reliability of predictions of the future evolution of the ozone layer over the Arctic depends on the degree of maturity of current chemical models of the polar ozone loss processes. Presentations during the session focused on comparisons of model calculations with empirically determined ozone losses.

R. Salawitch focused on model runs constrained by measurements of ClO and Cl2O2 by the ER-2 during the SOLVE/THESEO 2000 campaign. The model results were compared with ozone losses determined from the ER-2 ozone measurements. Using the current (preliminary) version of the ClO data, the model underestimates the observed ozone loss considerably for all ER-2 flights. However, taking into account expected revisions of the ClO data (as was discussed in halogen session) reduces the discrepancy and results in reasonable agreement between the model and the observations for most of the flights, perhaps with the exception of the missions early during the year.

J.-U. Gross reported box model calculations along trajectories used in Match ozone loss studies for the years 1992 and 1995. During January of both years the observed ozone losses along the trajectories were significantly larger than the model could reproduce. The discrepancy for 1992 is shown in Figure 3. Using the CLAMS CTM, he analysed whether the Match results inside the polar vortex may have been impacted by dynamical supply of ozone poor air from outside. He found that in January 1992 some mixing occurred, but that the estimated impact was by far too small to explain the discrepancy.

Figure 3: Ozone loss rates from Match (red) compared with model calculations (blue) along the air mass trajectories used in the Match study [Becker et al., Ozone loss rates in the Arctic stratosphere in the winter 1991/92: Model calculations compared with Match results, J. Geophys. Res., 1998]. The plot shows an underestimation of the observed ozone loss by up to date chemical models during January. Since Becker et al., 1998, this underestimation has been confirmed by independent studies that were reported at the Workshop and for all cold Arctic Januaries that have been studied so far (i.e. 1992, 1995, 1996, and 2000).
(For a better resolution of the images, please click on the plot or contact the SPARC Office)

T. Hanisco focused on the balance between active chlorine species and their passive reservoirs during the recovery period. Based on ER-2 observations and model studies he reported that the relevant processes are extremely sensitive to surface reactivity, calling into question how accurately they can be represented in models of the Arctic stratosphere.

Results from a multiannual integration of the SLIMCAT CTM were presented by M. Chipperfield. Using a microphysical scheme that requires the formation of ice particles before substantial denitrification can occur, SLIMCAT has not produced significant denitrification in any winter before 1999/2000 and has generally underestimated ozone loss during cold Arctic winters, with the exception of 1999/2000 [ B.-M. Sinnhuber]. In 1999/2000 a cold bias in the UKMO temperature fields led to unrealistically intensive formation of ice particles in the model, resulting in widespread denitrification and more realistic ozone loss. Using the same model G. Millard addressed the interannual variability of ozone loss and argued that in the Arctic the ClO-BrO cycle is the predominant loss cycle.

M. Rex showed that unrealistically high levels of active chlorine would be required to explain Match observations of ozone loss rates in cold Arctic Januaries with the known photochemistry. Based on a statistical analysis of Match data, he showed that the January ozone loss occurs most likely during short sunlit periods at high solar zenith angles. For March 1996, he found good agreement between the Match ozone loss rates and model calculations constrained by MLS CLO measurement, using version 5 data.

An overview over results from coupled chemistry-climate models were given in a contribution by J. Austin, presented by V. Eyring. The range of predictions from different models is broad. The initial concern raised by the GISS model that very low Arctic ozone levels can occur around the year 2020 due to increasing greenhouse gas concentrations was not confirmed by the majority of subsequent studies with other models. Some are showing a rapid recovery within the next two decades and even a positive impact of greenhouse gas emissions on Arctic ozone recovery.

The discussion in this session focused on the discrepancies between ozone loss observations and models. Two points were identified where major uncertainties in our understanding remain. First, the role of denitrification for Arctic ozone losses was generally acknowledged and uncertainties in current microphysical schemes are believed to contribute significantly to a general underestimation of the degree of ozone loss in cold Arctic winters. Second, Arctic ozone losses during cold Januaries appear to be larger than can be explained with current models. The January losses do not represent a large fraction of the overall ozone loss over the course of a cold Arctic winter and the impact of these uncertainties on bulk ozone losses is limited. However a full understanding of the processes is required as basis for reliable predictions of future ozone levels in the Arctic stratosphere.

A second area of discussion centred on the current status of coupled chemistry-climate models. The general view was that the impact of increasing greenhouse gas levels on Arctic ozone remains unclear and that a thorough validation of existing models is required to make progress. This should focus on the validation of processes, e.g. by validating the relation of chemical ozone losses to average PSC areas or the relation of Arctic temperatures to mid-latitude wave fluxes, rather than on the validation of bulk quantities like the total ozone column.


The Arctic Ozone Loss Workshop brought together members from different communities to address the uncertainties that limit our ability to ‘close’ the Arctic ozone budget. Several important conclusions were reached by consensus of the group. First, different techniques for quantifying chemical ozone loss in the Arctic vortex are providing a consistent picture of those losses. Second, the measured abundances of ClO and BrO appear to be sufficient to explain ozone losses during the period of most rapid change (late winter and spring). Finally, accurate temperature analyses and realistic microphysical schemes were viewed as being critical to the ability for models to reproduce the observed ozone losses and their interannual variability.

Another important conclusion reached at the Workshop was that ozone losses in mid-winter (i.e. at high solar zenith angles) are real and cannot be fully explained by standard halogen photochemistry. Although these ozone losses are typically small relative to losses later in the season, the fact that they cannot be explained calls into question our ability to predict future trends in Arctic ozone.

To close out the Workshop, the participants outlined a general strategy for future observations and process studies. It was agreed that continuous observations of O3, ClO, H2O, HNO3, temperature, and polar stratospheric clouds, preferably with large spatial coverage, will be critical for characterising the changes expected as earth’s climate changes and as chlorine loading decreases throughout the century. Process studies addressing nucleation of PSC particles and subsequent chlorine activation and denitrification will be crucial to constrain model parameterisations of particle microphysics, necessary for developing a credible predictive capability. New studies of early- and mid-winter ozone losses and the possible role of unknown halogen reactions will be important for reducing the probability of a ‘surprise’.

The group was optimistic that the tools and collective experience are available to address these needs, and that the ability to predict the future evolution of Arctic ozone should be the goal of future research efforts by this community. In particular, research programmes should focus on the response of Arctic ozone to changes in wave driving, source gas abundances (e.g. chlorine, bromine, methane, and H2O), radiative forcing, and denitrification.


Back to SPARC Newsletter 19 Homepage