ECSS2023-73, updated on 03 Mar 2023
https://doi.org/10.5194/ecss2023-73
11th European Conference on Severe Storms
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

A New Ensemble System at DWD for Seamless Short-Term Forecasts of Convective Precipitation

Martin Rempel, Markus Schultze, and Ulrich Blahak
Martin Rempel et al.
  • Deutscher Wetterdienst, Forschung und Entwicklung, Offenbach am Main, Germany (martin.rempel@dwd.de)
Reliable and accurate forecasts in the short-term range are essential for precise and consistent warnings that help to increase the lead time for decision makers in emergency services. In the current operational weather forecasting, these warnings are commonly based on nowcasting techniques and numerical weather prediction (NWP). Both forecast systems are able to provide valuable warnings guidance, albeit for different lead-time ranges. Especially in convective environments, the skill of nowcasting techniques based on Lagrangian persistence is often limited due to large dynamical uncertainties. On the other hand, NWP forecast quality may be affected by incorrect initial conditions, outdated boundary conditions or by model spin-up effects.
The ongoing DWD project SINFONY (Seamless INtegrated FOrecastiNg sYstem) developes an integrated short-term ensemble forecasting system on the convective scale. As a first step towards the combination of precipitation and NWP, both techniques have been further enhanced. The previously purely advective precipitation nowcasting was replaced by STEPS-DWD, providing ensemble extrapolations every 5min. It represents an adaption of the well-known STEPS (e.g. Seed 2003, Bowler et al., 2006). On the part of the NWP, the new rapid update cycle (RUC) of ICON-D2 provides hourly initialized ensemble forecasts running 8h ahead with a horizontal resolution of 2.2 x 2.2km². Consistent with the radar scanning strategy, output of precipitation variables and synthetic radar reflectivities is given every 5 minutes.
To condense information from both described forecast systems and to provide an improved basis also for hydrologic warnings, the combined ensemble forecasting system INTENSE (Integration of NWP Ensembles and Extrapolations) is introduced. INTENSE adapts the Bayesian combination approach according to Nerini et al., 2019 by utilizing the ensemble Kalman filter in a dimension-reduced space. It provides combined forecasts up to 6h ahead with a spatial and temporal resolution of 5min and 1 x 1km², respectively.
This contribution will give an overview about the adaption of the combination approach and will discuss additional modifications. Further, a verification study for the summer 2022 will be shown, highlighting the benefit of combined forecasts for several severe convective events over Germany. In addition, results of an evaluation carried out by forecasters in the same period will be presented.

How to cite: Rempel, M., Schultze, M., and Blahak, U.: A New Ensemble System at DWD for Seamless Short-Term Forecasts of Convective Precipitation, 11th European Conference on Severe Storms, Bucharest, Romania, 8–12 May 2023, ECSS2023-73, https://doi.org/10.5194/ecss2023-73, 2023.