Informal Seminar: On the post-processing of ECMWF forecasts at Météo-France

by Maxime Taillardat (Météo-France - CNRM UMR 3589)

Lecture Theatre

Lecture Theatre


Like other European meteorological centres, Météo-France makes an extensive use of ECMWF's forecast and products, from short lead times to the extended range, whether for decision support of human forecasters or in the production of automated weather reports. This talk will provide a quick overview of the ECMWF products used at Météo-France and products derived from ECMWF forecasts, such as MOS products or automated charts. In particular, the current statistical post-processing chain will be described. It is based on the development of several post-processing algorithms, from the simplest to the most sophisticated, deployed according to a computing cost / performance ratio and the amount of training data available. The existing ECMWF post-processed products coming from deterministic MOS, ensemble calibration and/or downscaling will be presented together with blending strategies for deterministic and ensemble forecasts. At last, future challenges in statistical post-processing will be discussed such as the rise of AI techniques, the calibration in space and time, extreme events, and verification issues.