16B.1 Analysis of Radar QPE for the 2021 Regional Flooding Events in Germany, Belgium, Luxembourg, and the Netherlands

Friday, 1 September 2023: 8:00 AM
Great Lakes A (Hyatt Regency Minneapolis)
Edouard Goudenhoofdt, RMI, Brussels, Belgium; and H. Leijnse, L. Bogerd, J. Y. Chen, L. Delobbe, M. Gottschalk, T. Hengstebeck, L. Mathias, K. Mühlbauer, A. Overeem, S. Trömel, T. Vlemmix, E. Weigl, and T. Winterrath

The 2021 regional flooding events in Germany, Belgium, Luxembourg, and the Netherlands caused severe damages and many fatal casualties. Flooding was triggered by a widespread, long-lasting stratiform event with some embedded convection that passed over the area and stalled for two days. These storms caused floods through both locally extreme-intensity precipitation and extreme precipitation accumulation over 2-3 days. National operational radar QPE products of Germany, Belgium, and the Netherlands all severely underestimated the precipitation. We aim to investigate the main causes of this underestimation, and to suggest robust correction algorithms for them.

We have started a collaborative effort to bring together all data relevant to the event, in commonly agreed data formats. This includes dual-polarization C-band weather radar data from national radar networks, dual-pol X-band and vertically-pointing K-band (MRR) radar data from the University of Bonn, rain gauges from national and regional networks, and disdrometer data. These data are used to investigate which processes were not properly captured by the different national QPE products. We find that Doppler notch filtering (especially near the zero-isodop), attenuation, vertical variation of precipitation related to warm rain processes and orographic enhancement, and assumptions on the raindrop size distribution (DSD; and related Z-R relations) play major roles in the underestimation for this case.

Several methods to improve the precipitation products based on information that would be available in real time are tested and evaluated. Attenuation correction based on differential phase and KDP-based rainfall intensity retrieval are an important part of this. Intelligent combination of data from different elevation scans and from different radars are also seen to improve QPE. Merging with (near-real-time) rain gauge accumulations is seen to help mitigate the effect of both precipitation enhancement below the lowest radar scans and assumptions about the DSD. However, this heavily depends on the spatial density of rain gauges that are available in real time.

We present results of our investigations into the sources of underestimation, and how the different methods aimed at improving the QPE products contribute to more accurate precipitation estimates. And we provide recommendations on how we can learn more from events like this in the collaborative framework we have set up for this.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner