The Global Extremes Digital Twin of Destination Earth: successes and challenges

At the end of the first phase of the Destination Earth initiative in May 2024, ECMWF will deliver a prototype of the global component of the Weather-Induced Extremes Digital Twin (or Global Extremes DT). The Global Extremes DT will monitor worldwide extreme weather events up to 5 days ahead and at an atmospheric resolution of 5 km. Furthermore, it incorporates two impact sector models: the CaMa-Flood river routing model for predicting flood risk, and a flexible aerosol scheme that monitors selected aerosol species, contributing to air quality assessment. Since August 2023, a daily forecast is performed with the prototype Global Extremes DT on the ECMWF's HPC Atos. In this presentation, we will summarise the work done during the first phase of Destination Earth and revisit some successes and challenges encountered in predicting extreme weather events.

A large selection of extreme weather cases has been analysed to demonstrate the added value of the high-resolution DT over the ECMWF operational forecast. Clear improvements were found for near-surface fields in regions of complex terrain, the intensification of tropical cyclones, the magnitude of orographic precipitation and subsequent flood events. Yet, we also find that standard NWP scores are not improved readily when the horizontal resolution is increased, and specific developments had to be made to adapt the physics of the model and exploit the full benefit of high-resolution. Although we find that non-hydrostatic effects do not matter up to a resolution of 2.8 km, and thus do not justify the extra cost of a non-hydrostatic dynamical core, an exception is strong subtropical jet regimes over high orography. In that case, the orographic gravity waves are not handled correctly by the hydrostatic dynamical core. As a result, the model’s mean orography needed special filtering at small scales and the model timestep had to be reduced. The degradation of NWP scores at kilometre-scale also prompts the need to find new ways to assess the improvement of physical realism.

Finally, we will give an overview of the current developments at ECMWF that will be part of future versions of the global extremes DT, such as a new prognostic TKE turbulent scheme, modified settings for atmospheric convection and a revised orographic drag parameterization.