The workshop is co-sponsored by ECMWF, WMO/WGNE, WMO/THORPEX and WCRP

Traditionally, parametrisations in numerical weather and climate prediction models have been formulated deterministically. Estimates of the forecast uncertainty that can be attributed to model imperfections have previously been estimated using quasi-independent models, quasi-independent parametrisations within a given dynamical core, and/or quasi-random parameter perturbations within a given model framework. In recent years an alternative approach is emerging, whereby the parametrisations (and hence the computational representations of weather and climate) are partially stochastic. The purpose of this meeting is to discuss the origins of model uncertainty in terms of physical processes, to compare the emerging methods for representing model uncertainty with the more "traditional" methods, and to make recommendations on how this important area of model development should proceed. The meeting will bring together researchers from a range of disciplines.

Expected attendees will include scientists from universities, labs, and weather and climate prediction facilities. To the extent possible, we also hope some graduate students and post-doctoral fellows with an interest in the subject will attend. There will some opportunity for poster presentations.

En savoir plus