Diagnosing, representing, and reducing Earth System Model uncertainty for weather and climate predictions

Logistics

We sent out an information email on Wednesday 18 September with further information for workshop attendees. Please get in touch on modeluncertaintyoxford@gmail.com if you did not receive this email.

Frustratingly, some pages on the webpage seem to be slow to update for some users. Our apologies. The current version of the schedule is 3.3:

https://oxfordmodeluncertainty.web.ox.ac.uk/sitefiles/schedule-draft-03....

Aims

Future predictions of the Earth system are uncertain, whether spanning a few days or many decades.

A key source of uncertainty across all these timescales are the approximations made when building the Earth System Model used to make the prediction. This is called model uncertainty. 

This workshop will bring together different communities interested in model uncertainty, including but not limited to: those working on weather prediction through climate timescales; those focused on physical parametrisations, the dynamical core, or their coupling and interaction; those using high-resolution km-scale models through complex Earth System Models; whether limited area or global. 

We welcome papers on quantifying, understanding and/or representing model errors and model uncertainty, including through multi-model ensembles, parameter perturbation experiments, or stochastic approaches, among others. Statistical approaches including machine learning are also of interest.

The workshop will include invited talks, contributed talks, and posters. We will hold breakout groups to foster deeper discussion on the key workshop themes. We plan for the workshop to be partially hybrid - while all presentations will be in person, we will livestream talks and facilitate online Q&A. For more details on how you can participate, please see the 'Registration' page.

Key workshop themes

• Representing model uncertainty including the use of ensembles; stochastic parametrisations; multi-model and perturbed parameter approaches; model error and uncertainty in data assimilation

• Understanding and quantifying model error; observational constraints; 

• Model uncertainty across model hierarchies, across temporal scales, and across spatial scales; the use of unified or seamless approaches to improve and constrain predictions; 

• Resolution issue: the development of scale-aware parametrisation schemes; parametrisation across the grey-zone; km-scale modeling

• The role and use of machine learning (ML) in all the above; model uncertainty in an ML context

Confirmed Invited Speakers

V. ​Balaji (Princeton / Schmidt Sciences)

Ben Booth (UK Met Office)

Christoph Schär (ETH Zurich)

Antje Weisheimer (ECMWF / University of Oxford)

Matthew Willson (Google DeepMind)

Franç​​​ois Bouttier (Météo France)​​​