Report on the SPARC/IOC Workshop on Understanding Ozone Trends,
Department of Meteorology, University of Maryland, College Park, USA, March 7-9, 2001

N. R. P. Harris, EORCU, Cambridge, UK (Neil.Harris@ozone-sec.ch.cam.ac.uk)
R. D. Hudson, University of Maryland, College Park, USA ( hudson@atmos.umd.edu)

Background

In the past few years, continued observations and recent analyses have raised a number of questions about how well past ozone changes can be explained in quantitative terms. In particular the importance of changes in halogen loading, stratospheric aerosol loading, atmospheric dynamics radiation and temperature have been assessed in a number of ways. However there is no general consensus on the relative contributions of these factors to past ozone changes or on how to assess them in a consistent manner. This is an important topic as the peak stratospheric halogen loading is passed. Such issues will be discussed in the next (i.e. 2002) WMO-UNEP Scientific Assessment of Ozone Depletion.
Accordingly, a workshop on the understanding of ozone trends was organised to stimulate discussion and relevant work ahead of the preparation of the 2002 assessment and to see what issues will require research over the longer term. This was done at the initiative of the SPARC Ozone Trend working group and the International Ozone Commission (IOC) in close conjunction with the co-chairs of the Montreal Protocol Scientific Assessment Panel and with WMO/GAW. Over 40 leading scientists in this field attended the workshop, including members of other SPARC working groups. The programme was balanced between presentations of new results and discussion of the issues raised.

Scientific Issues

There are a number of important scientific issues:

• The statistical model of ozone trends used in recent assessments has not allowed for long-term dynamical changes other than those related to the QBO and solar cycle. Nor has it yet incorporated the fact that the chlorine loading has stopped increasing and that the ozone-depleting efficiency of chlorine compounds is strongly modulated by stratospheric aerosols. The statistical approaches used to calculate ozone trends may need to be updated and developed in sophistication to address today’s questions.
• There has been increasing evidence from published studies and work in progress that long-term dynamical changes (possibly related to climate change) may have been responsible for a significant fraction of the ozone trends derived from measurements at mid-latitudes. The studies vary from combined analyses of ozone and meteorological data through increasingly sophisticated 2D model studies using time-varying aerosols, etc. to 3D model runs using real winds. Estimates for the contribution for the dynamical contribution to the observed ozone trends have varied from zero to 100%. Simple-minded addition of all the current dynamical and chemical explanations of the mid-latitude trends would account for nearly twice the observed trend. Sound scientific approaches need to be developed so that these influences on ozone can be allowed for consistently in the statistical models used in the assessment.
  
• There are some signs of a change of behaviour, most obvious in the tropics, in the TOMS measurements of ozone since 1998 and a further shift in 2000. While discussion of issues related to the data quality of the various observing systems were not a major part of this workshop, attention was given to identifying whether such features in the observational record are real and to updating the estimates of the instrumental uncertainties.
  

Aims

The main aims of the workshop were to

• Identify the major current issues concerning ozone trends,
• Improve quantification of the contributions and uncertainties of the chemical and dynamical mechanisms to observed ozone trends, particularly at mid-latitudes,
• Identify how to assess the consistency of these proposed contributions.

Success in meeting these aims has resulted in significant progress in the preparations for the forthcoming 2002 WMO-UNEP assessment. Steps that will assist the assessment process (further work, comparisons, publication, etc.) were explicitly discussed at the workshop.
In addition, although not a primary aim, there was some discussion on how the approaches that can be developed to improve our understanding of past changes could be used to identify the recovery of the ozone layer as a result of actions taken under the Montreal Protocol.

Discussion and findings

A. Data quality and availability

The overall quality and availability of ozone data are in reasonably good shape. Clarification is needed however on the possible satellite bias between the Northern and Southern hemispheres and some apparent degradation in EP-TOMS in 2000. Several composite data sets for total ozone, in which the instrument to instrument biases have been addressed, have been prepared. The aim is to inter-compare these composite data sets and to have identified and understood discrepancies by December 2001, in time for the assessment. The need for continued high quality ozone data from ground and space was clearly recognised.

B. Chemical Influence

Current estimates of chemical loss of ozone, which come principally from 2D models, have changed little since the 1998 WMO-UNEP assessment, the biggest change resulting from improved reaction rates for NOx chemistry. The 2D models can reproduce most of the northern mid-latitude ozone trends. In particular, several 2D models have simulated the observed drop in ozone following the Mt Pinatubo eruption in 1991, as well as the more recent higher values, supporting the link between chlorine chemistry and aerosol loading. The agreement between measurements and models does not appear to be as good for the Southern Hemisphere. It has proven hard to accurately quantify the impact of polar chemical processes at mid-latitudes (a certain influence on the mid-latitude trends) from measurements and from models. The high altitude loss should be the simplest to model.

C. Dynamical Influence

There was much discussion on the influence of dynamics on ozone abundances and trends. Ozone trends in the lowermost stratosphere are particularly sensitive to changes in dynamics, whether changes in Rossby wave events, in the frequency of low ozone events (mini-holes and laminae) or changes in principally tropospheric phenomena such as the North Atlantic Oscillation / Arctic Oscillation. Diagnosing the links between these undoubtedly coupled phenomena is hard, but it is necessary in order to avoid ’double counting’ when calculating ozone trends.
In the long term, the magnitude of the dynamical influence on ozone trends depends critically on whether the short term relations between dynamics and ozone stay constant (as currently assumed) or vary with time as the overall atmospheric system responds to the original forcing (whatever its origin). As a result, the dynamical influence depends on the time period considered. The current estimates of the influence are likely to be upper limits. It is important not to simply add ozone trends derived from individual statistical studies looking at the influence of one particular dynamical process, as the various dynamical processes in the atmosphere are clearly related.
It is unclear which dynamical proxies should be used in statistical models. A better understanding of which dynamical processes affect the ozone is clearly needed, as well as any radiative feedbacks or radiative forcings from the changes in stratospheric water vapour or greenhouse gases. Simple 3D CTMs, driven by assimilated winds, show a large inter-annual dynamical influence on ozone trends and this will affect the estimate of ozone trends. It is hard to attribute causality in models; i.e. what is a feedback and what is a driving force? Improved understanding of ozone-climate interactions and past temperature trends is needed for improved characterisation of past ozone changes.

D. Regimes

Another approach, which takes account of atmospheric motions, is to consider the atmosphere as a number of regimes with boundaries chosen according to dynamic criteria rather than simple geographic criteria, which in ozone trend studies have typically been latitude bands. New work was presented that uses the meteorological tropopause fronts to separate total ozone into regimes. Narrow distributions of total ozone were found if the regions close to the fronts were excluded. Within the regimes, small trends in total ozone were observed. Changes in the relative contributions of the tropical, mid-latitude and polar regimes to the overall trends at mid-latitudes were also reported. Trends in the lowermost stratosphere, calculated relative to the tropopause height, have been used to distinguish stratospheric and tropospheric effects. In addition, polar ozone trends have been investigated using equivalent latitudes on a number of isentropic surfaces in the lower stratosphere.

E. Use of ‘Fingerprints’

There is a need to identify, consistently, the spatial (latitude, longitude and altitude) and temporal signatures expected in ozone as a result of the various chemical and dynamical processes. Attempts should be made to quantify ozone loss, temperature, chlorine, etc. at 40 km using adapted climate ‘fingerprinting’ techniques. At lower altitudes one can probably only make qualitative estimates at present. Wherever possible, external variables, such as water vapour, methane, hydrogen chloride, temperature, etc. should be used.

F. Statistical models

Much of the discussion on this subject centred on which proxy variables should be used to describe the dynamical influence and the halogen-induced ozone loss. Ideally such proxies should be independent (orthogonal), but this may not be possible given that the atmospheric processes involved are coupled. This issue remains an important one to resolve.
The development of tools to investigate the turnaround resulting from decreased halogen loading is progressing, and it currently seems that any turnaround will not become observable until at least 2010. An improved understanding of the processes affecting ozone trends is an essential element of this work. Most of the current statistical effort is being devoted to the further development of time series approaches rather than the introduction of different statistical techniques.

The way forward

The main areas of activity in the coming year will be the continuation of the on-going research projects and the preparation of the 2002 WMO-UNEP report. In order to facilitate general discussion and information flow on this issue, the IOC will host a joint IOC/SPARC web site on understanding ozone trends. It is anticipated that a further joint initiative will be undertaken in the second half of 2002 which will promote progress on the issues which are still outstanding after the WMO-UNEP report’s considerations.
As stated above, the understanding of the chemical depletion of ozone has not changed much since the 1998 WMO-UNEP report, and those advances which have occurred have done so principally as a result of revisions of laboratory data. Recently there has been more work using chemical composition measurements to provide constraints on the chemical loss mechanisms in the lower stratosphere, and this work needs to continue. A thorough evaluation of high altitude ozone trends taking into account all the known or likely changes in chemical composition, temperature and possibly large-scale circulation in that region is still needed.

It is now clear that the effect on ozone of the decadal changes in dynamics needs to be considered in the calculation of ozone trends. What is currently less clear is how to evaluate the magnitude of their effect on ozone quantitatively and how to separate purely dynamical changes from each other and from those induced by the ozone changes themselves. A more comprehensive account of all the dynamical processes needs to be made simultaneously on a hemispheric basis. The very recent work done where total ozone is separated according to meteorological regimes at the tropopause provoked a great deal of interest, but how the results relate to the previous view of mid-latitude ozone depletion remains a puzzle.
Developments in 3D models have meant that these have started to be used in serious trend studies and they offer a new way to assess the relative importance of chemistry and dynamics. However there is still a lot of work to be done in this area in order to get realistic descriptions of chemical processes, to develop appropriate diagnostics to distinguish cause from effect and to evaluate the consistency with the current 2D models.
Overall, further developments are likely to be made along the general principles governing the evolution of the overall SPARC programme. These will involve progress toward a more integrated understanding of all the changes which have occurred in the stratosphere (water, temperature, radiation, dynamics as well as chemical composition including halogen and aerosol loading). However a word of caution is needed here as the uncertainties associated with the observed changes may preclude an exact interpretation.

Acknowledgements

We wish to thank the participants of the meeting for their enthusiasm during the workshop and for their assistance in the preparation of this summary. We thank SPARC, the University of Maryland, the EC Research DG (through the CRUSOE concerted action – EVK2-1999-00252) and the UK DETR Global Atmospheres Division for their support in the organisation of the workshop, as well as the many other funding agencies who helped support individuals to attend.

Attendees

A-L. Ajavon, D. Albritton, D. Balis, G. Bodeker, R. Bojkov, D. Cunnold, V. Fioletov, E. Fleming, L. Flynn, P. Forster, S. Godin, N.R.P. Harris, S. Hollandsworth, L. Hood, R.D. Hudson, I. Isaksen, J. Kerr, J. Logan, R. McPeters, G. Mégie, A.J. Miller, H. Nakane, M. Newchurch, P. Newman, S. Oltmans, R. Portmann, M. Proffitt, J.A. Pyle, V. Ramaswamy, W. Randel, S. Reid, G. Reinsel, N. Sabogal, S. Solomon, J. Staehelin, R. Stolarski, O. Uchino, B. Weatherhead, M. Weber, D. Wuebbles, J. Zawodny, C. Zerefos

Back to SPARC Newsletter 17 Homepage