The WMO/WCRP Working Group on Numerical Experimentation (WGNE) organises a hybrid workshop on systematic errors in weather and climate models, hosted by the European Centre for Medium-Range Weather Forecasts (ECMWF) on 31 October - 4 November 2022.
The workshop brought together a wide range of experts on simulating the Earth System including atmosphere, ocean, waves, land-surface, atmospheric composition, and associated disciplines to advance the understanding of systematic simulation errors at all timescales. A particular emphasis was given to identifying errors in complex coupled systems and to understand their root causes. Progress in diagnosing and addressing systematic errors using a wide range of tools ranging from classical methods to advanced technologies such as data assimilation and machine learning / AI was documented. The workshop encouraged an active discussion on relative merits of active development of physical models and parametrisations to address systematic errors versus bias correction methods.
The workshop reviewed recent progress made on the atmospheric systematic error priorities identified from the 5th Workshop on Systematic Errors, while also expanding focus to coupled systems. The workshop was broadly organized around the following themes:
- Clouds and precipitation: convective precipitation; sfc-fluxes and diurnal cycle; cloud microphysics; representation of low clouds, especially at high latitudes; uncertainty representation; resolved/unresolved convection; precipitation over orography; convective organisation.
- Atmosphere-land-ocean-cryosphere interactions: sfc fluxes and diurnal cycle; surface drag; soil/vegetation/land-use representation; stable boundary layer issues; fog-stratus simulation; impact of coupled modeling to ocean/sea-ice/waves.
- (sub-)tropical circulations: Tropical cyclones, MJO, QBO, double ITCZ and ENSO biases, ocean and wave coupling, role and biases over the maritime continent, tropics and mid-latitude coupling.
- Stratosphere-Troposphere interactions: role of stratospheric biases on longer term predictability; stratospheric bias correction and impact on predictability; atmospheric composition and long-lived tracers.
- Machine learning/AI and data assimilation: novel techniques to diagnose, measure or resolve systematic errors.
- Quantifying uncertainty: spread-error relationships; identifying role and contribution of physical processes to uncertainty characterization ; ensemble and hindcast strategies to identify extremes ; use of multi-model ensembles to identify systematic errors; stochastic representations of model uncertainty.
- Challenges and surprises in simulating the climate system: improvements and new errors with eddy resolving / deep convection resolving simulations, errors due to vertical/horizontal resolution imbalances, shifts in statistical analyses due to resolved vs parametrised processes, systematic errors due to changes in convective organisation
The format of the workshop facilitated online and physical participation and followed the previous successful workshop in Montreal in 2017
There was a single session of oral presentations (no parallel sessions) from selected abstracts plus keynote talks of solicited speakers.
The majority of presentations were in the form of posters which were discussed in dedicated physical sessions and a dedicated online session.