The Multi-Faceted ECMWF Approach to Forecasting Hazardous Weather

ECMWF utilises its supercomputer resource to generate bewildering volumes of forecast and re-forecast data every day. To convert these into meaningful products related to hazardous weather, in a way that will directly benefit society, certainly requires technical expertise, but above all requires a comprehensive understanding of the meteorology of different phenomena, and of the ability of modelling systems run at different resolutions to represent them. The product range ECMWF provides to users has been carefully expanded over the years with these aspects in mind, and now provides useful guidance on phenomena that can be accurately represented with current generation global models (such as stratiform rainfall around extra-tropical cyclones), as well as those that cannot (such as extreme convective hazards). This presentation will provide examples of recent and ongoing developments in both categories, for medium range timescale, to include “new” phenomena such as freezing rain, fog, lightning and extreme localised rainfall. It will highlight also how using the powerful concept of the Extreme Forecast Index (EFI), which is based on comparison with a same-model reforecast climatology, can circumvent resolution-related limitations, and help address the need for impact predictions. There will also be reference to aspects where international collaboration is proving very beneficial (e.g. in the hydrological sphere). Areas where the challenges for global modelling remain particularly acute, such as lake effect snow, will be discussed.

The ECMWF 10-year strategy envisages pushing the limits of predictive skill further into the future, and in conjunction, following a successful proof-of-concept verification activity, we are now developing EFI-style products for the extended ranges, to run out to week 4 or even week 6. This will be discussed too, referencing also the need to manage expectations.