Group seminar on 13. April, 14:15 CET

Scientific concepts underpinning the design of probabilistic forecasting systems for weather and climate

This lecture will introduce the seamless probabilistic forecasting systems used at ECMWF to produce
forecast of weather and climate from days up to one year ahead. We will discuss the first and second
kinds of predictability (initial value and boundary forcing), and the use of ensembles to deal with the
probabilistic nature of the prediction problem. A few techniques to extract information from the
ensemble, which include reducing the dimensionality of the system will be illustrated. Finally, we
will discuss different forecasting system strategies to deal with model error.