The GLObal-to-Regional ICON (GLORI) Digital Twin: towards hectometric scale predictions for high-impact weather

The Global-to-Regional ICON (GLORI) Digital Twin is a configurable on-demand global-to-regional short-range high-resolution digital twin based on ICON. It is developed in a tri-lateral cooperation between Germany, Italy and Switzerland.

GLORI provides short-range global predictions down to the storm-scale (~3 km horizontal) and on-demand high resolution (~ 500 m) predictions for selected regions, like the Alpine domain or the Italian peninsula. It includes an uncertainty estimation through ensemble forecasts for global and regional scales. The data assimilation system is ensemble-based, both for the global and the regional components. The GLORI Digital Twin aims at providing forecasts down to the application level, for a range of use cases including flood forecasting, urban heat island and urban flooding events, mineral dust predictions for energy applications and pollen predictions.

Moving to higher resolution requires improvements both in the model and in the data assimilation system. We test the ICON model in complex topography and highlight its behaviour in dependence of conditions like stable boundary layer, flow interacting with the orography, convection development, different soil textures and urban areas. In that, GLORI can also be seen as testing environment for the development of hectometric scale modeling. The research focuses also on data assimilation at higher resolution, both for the global and for the regional runs, and on the usage of high-resolution observations. The assimilation of the radar data of the three partner countries is tested over the Alpine domain, aiming at the improvement of the prediction of convective events. Dedicated studies focus on direct assimilation at 1 km resolution. This goal demands a rigorous evaluation of the entire assimilation workflow, including observation thinning, averaging strategies, localization, and observation error quantification. The impact of performing data assimilation at 2 km resolution with nesting at 1 km is then compared with direct assimilation in the 1km domain. The performance of the Digital Twin is assessed on high-impact weather events, considering in particular convective development leading to severe weather and the recent flood events.