ML for weather prediction at Météo-France : current status and future plans

Applications of Machine Learning (ML) in the different stages of weather forecasting have considerably developed recently. Such progress is likely to change the landscape and offer new perspectives to speed up and improve forecast performances, at different spatio-temporal scales. In this context, Météo-France engaged more actively in this new area of research, with the objective to further explore the capabilities and opportunities of ML for operational forecasting. Major ongoing projects include ML to significantly enhance the size of convective-scale ensemble forecasts, high-resolution statistical downscaling and the development of data-driven kilometre-scale forecasting systems. Early results will be presented and our short-term roadmap will be discussed.