Tornado in Czech Republic – An ICON-RUC case study

The German Meteorological Service (DWD) aims to enhance its ability to provide warnings on shorter notice (e.g. of extreme convective events) and therefore implements a rapid update cycle (RUC). This system produces more frequent and faster available numerical ensemble forecasts compared to the standard ICON-D2 short range numerical weather prediction (SRNWP) forecast procedure. The RUC’s focus is on lead times of up to 12 hours.

The RUC’s capability in predicting extreme convective events is tested by applying its setup to re-forecast the violent tornado case near Hodonin in the Czech Republic in June 2021. As this area is outside of the ICON-D2 domain, the model domain was shifted. While several German radar stations could not contribute to the new domain, 3D radar reflectivities and radial winds from several Czech and Slovakian radars were used for this case study. The assimilation cycle is started from the ICON-EU state the day before the event and hourly ensemble forecasts are started before the actual event. In addition, we included a nested domain, which uses a resolution of ~1km to evaluate the impact on the atmospheric parameters and dynamics.

We present results of the simulation of the atmospheric condition as well as tracks and dynamics of the developing convective systems with respect to different lead times and horizontal resolutions. As expected, there is an increase in forecast accuracy of the convective event with decreasing lead time. With respect to the different horizontal resolution, amplitudes of wind speed and updraft helicity are larger for the 1-km domain. Further, the rotation signature of the simulated tornadic supercell (e.g. radial wind) is comparable to the actually observed one.