A Workshop on ensemble prediction was held at ECMWF on 7-9 November 2007. The workshop aimed to review the most recent advances in ensemble techniques applied to data-assimilation and forecast systems for predictions ranging from days, through months and seasons, to multi-annual time-scales.
Topics discussed included:
- Representation of initial uncertainties (ensemble data assimilation, ensemble transform Kalman filter, bred vectors, singular vectors etc);
- Representation of model uncertainties (multi-model ensembles, perturbed parameter ensembles, stochastic parametrisation etc);
- Validation and calibration methods;
- Applications of ensemble forecasts. Results of both theoretical and practical research, using both global and limited area models, were included in the presentations.
Lessons drawn from intercomparisons of results on different time-scales, as encouraged in the World Climate Research Programme's Strategic Framework on seamless prediction, were focus of discussion at the workshop.
Presentations given at the workshop are available at: https://www.ecmwf.int/en/learning/workshops-and-seminars/past-workshops/.... A summary of the working group discussions and recommendations will be published in the workshop proceedings.