Process-oriented validation of
coupled chemistry-climate models


Brief report on the Workshop on Process-Oriented Validation of Coupled Chemistry-Climate Models
to be published in the SPARC newsletter No. 22 (January 2004)

Veronika Eyring, DLR-Institiut für Physik der Atmosphäre, Germany (
Neil Harris, European Ozone Research Coordinating Unit, UK (
Markus Rex, Alfred-Wegener Institut-Potsdam, Germany (
Theodore G. Shepherd, University of Toronto, Canada (
With contributions from D. Fahey, J. Austin, M. Dameris, H. Graf, T. Nagashima, and B. Santer

A workshop was held on "Process-oriented validation of coupled chemistry-climate models" on November 17-19, 2003 in Garmisch-Partenkirchen/Grainau, Germany. Approximately 80 participants attended from Europe, the United States, Canada, Japan, and New Zealand. The workshop was held under the auspices of the Institute for Atmospheric Physics of the German Aerospace Center (DLR), the EU research cluster OCLI (Ozone CLimate Interactions), and SPARC. The meeting was hosted by the DLR. Conveners of the workshop were Dr. Veronika Eyring, Dr. Neil Harris, Dr. Markus Rex and Prof. Theodore Shepherd, with additional support from the programme committee.

The workshop was motivated by the need to evaluate the skill of coupled chemistry-climate models (CCMs) to predict the future state of the ozone layer. Providing accurate and reliable predictions of future changes in stratospheric ozone is of central importance in climate studies. Simulating the interaction between chemistry and climate is of particular importance, because continued increases in greenhouse gases and a slow decrease in halogen loading are expected, which both influence the abundance of stratosphere ozone. In recent years CCMs with different levels of complexity have been developed. They produce a wide range of results concerning the timing and extent of ozone layer recovery [WMO, 2003]. This outcome has created a need to identify the main dynamical, chemical, and physical processes that determine the long term behaviour of ozone in the models and to validate these processes by comparison with observations and other models.

Unlike chemical transport models (CTMs), which are constrained to follow the meteorology of a particular year and can therefore be directly compared with measurements on a day-to-day level, CCMs simulate a climate that at best only bears a statistical relationship to the real atmosphere. As a result, validation of stratospheric CCMs presents particular challenges. First, it is important to separate errors in model chemistry from errors in model dynamics and radiation. For example, a temperature bias would lead to an incorrect prediction of polar ozone, even if a model's chemistry were correct. Second, natural dynamical variability means that a comparison of model results with measurements must be performed in a statistical manner. This is problematic, because it appears to take many decades to define a robust stratospheric climatology, especially in the Arctic winter. While tropospheric climate models can be validated, in part, by their ability to reproduce the climate record over the 20th century, the paucity of stratospheric climate data prior to the satellite era (post 1979) severely restricts such possibilities for validating stratospheric ozone.

For these reasons, validation of CCMs needs a process-oriented basis to complement the standard comparisons of model and observed climatologies. By focussing on processes, models can be more directly compared with measurements. Furthermore, natural variability becomes an aid rather than an obstacle because it allows one to explore parameter space and, thereby, more readily identify cause and effect within a model. In the context of stratospheric GCMs (i.e. without chemistry), process-oriented validation represents the level II tasks within the GCM-Reality Intercomparison Project for SPARC (GRIPS) [Pawson et al., Bull. Am. Meteorol. Soc., 81, 781-796, 2000]. The recent WMO/UNEP Assessment contained a first attempt at process-oriented validation of CCMs [Austin et al., Atmos. Chem. Phys., 3, 1-27, 2003]. The purpose of this workshop was to build on this foundation and develop a systematic, long-term approach.

The workshop brought together members of the CCM and CTM communities, as well as various measurement groups, to develop a list of key processes and to identify specific diagnostics and datasets that could be used to validate those processes. The group also included those experienced with model validation activities such as GRIPS, EU-TRADEOFF, NASA Models and Measurements-II initiative, and the Program for Climate Model Diagnosis and Intercomparison (PCMDI). The workshop was structured around six major topics: Transport Characteristics, Stratospheric Dynamics, Stratospheric Chemistry and Aerosols, Tropical Tropopause Layer including the Upper Troposphere and Lower Stratosphere, Tropospheric Forcing, Radiative Transfer and Balance. Presentations in each topic began with an overview talk, were followed by two to four solicited shorter oral presentations highlighting certain specific issues, and then were completed with extended discussion. Each topic included contributed poster presentations. Rapporteurs provided a written summary of the discussion in each session.

The lasting impact and the full benefit from the workshop will come from the concerted validation activity that will be based on the results of the meeting. This activity will unfold over the next couple of years and needs the support of a broad community. It is important that the validation procedures defined for this activity are accepted and valued by all participants in this joint exercise. With the help of the theme speakers and the rapporteurs, the program committee has compiled a list of key validation processes, associated diagnostics, and relevant datasets, that is based on the contributions and discussions at the workshop. For the next few months this list will be open for discussion by the whole community, including all interested parties unable to attend the workshop. During this process the list of diagnostics and datasets will fully evolve. The list in its final form, a more comprehensive report of the workshop, and the plans for the coming years will be published in the next SPARC Newsletter. We invite all interested parties to actively contribute to the discussion process and to the joint activity that will be defined as a result of it. The evolving list with diagnostics, developing ideas and a list of contacts for individual aspects of the validation activity can be found on the website of the workshop at