2.1 General context
The Hanseatic City of Hamburg is situated at the mouth of the Elbe river, in the influence range of the North Sea tides which may be even increased in amplitude by the river mouth. At the same time, drainage of natural rivers and of the sewer system become critical with very high tides so that a risk of inland flooding exists. This was the base scenario of the StucK project “Long term drainage management of tide-influenced coastal urban areas with consideration of climate change” (2016–2019), funded by the Federal Ministry of Education and Research in Germany, with the objective to improve the current warning situation in view of the tidal context. The results have been implemented in the WaBiHa system (Warndienst Binnenhochwasser Hamburg – warning service for fluvial floods in Hamburg) and are increasingly being transferred to other application sites.
2.2 Precipitation module
The basis of the precipitation nowcasts are
-
Radar data from 4 German Weather Service (DWD) radar stations: Boostedt, Rostock, Hanover and Borkum. The basic radar product is a polar measurement (Plan Position Indicator – PPI) with a spatial resolution of 250 m x 1° and a 5-minute time step.
-
Rain gauge measurements of 400 stations (DWD in Northern Germany with time step between 5 minutes and 1 hour)
The radar data are processed and corrected with the software SCOUT (hydro & meteo 2009), using numerous filter and correction methods (Jasper-Tönnies & Jessen, 2014). The rain gauge measurements are continuously used for the adjustment of radar measured rainfall sums, based on data of the past 3 hours. A composite with a resolution of 1 km x 1 km is produced every 5 min from the data of the four radar stations. An example of the adjusted composite is shown in Fig. 1.
The radar composites of the last 30 min serve as input for calculating nowcasts with a lead time of 1–3 h. The nowcast method is based on a cell-tracking algorithm, tracking rain cells with a minimum size of 20 grid points above a reflectivity threshold. Cells are tracked based on their position and size, and displacement vectors are determined. These vectors are used to approximate the 2D advection field. A Semi-Lagrange scheme is used to forecast the advection of the rainfall field. Additionally, growth and decay of rain cells are extrapolated. The method is a further development of the nowcast described by Tessendorf and Einfalt (2012).
The ensembles are created through variation of the individual cell properties. Thus, the past development of the cells’ properties as well as their uncertainties are determining the degree of variation for the ensemble.
A joint evaluation of the forecast quality of the ensemble nowcasts, the deterministic COSMO-DE, and the ensemble forecasts COSMO-DE-EPS was performed. The ensemble nowcasts and COSMO-DE-EPS outperform the deterministic forecast (Jasper-Tönnies et al., 2018).
2.3 Real-Time warning
The real-time warning system WaBiHa combines the current warning level at 39 water level gauges with the warning levels from the precipitation measurements and the forecasts. The three independently determined warning levels are used to derive a general warning level, for each of the water level gauge specific subcatchments. The real time warning level is publicly displayed on the WaBiHa platform www.wabiha.de and changes of warning levels are emailed to relevant users within the city administration.
The updated warning system WaBiHa with ensemble nowcast based warnings has been operational since 2019. An evaluation of the highest 35 events in a 6-month-period with exceedances of warning levels at the water level gauges revealed that in 66% of the events a correct warning was issued based on the precipitation forecasts at least 6 hours ahead. The overall hit rate was 77% while the false alarm ratio in the same time period was 40%. Misses of the warning system were partly due to missing incoming data. These results represent a major step forward compared to the previous warning system which was solely based on the deterministic COSMO-DE forecast.
2.4 Further updates and consequences
The public WaBiHa portal has had a relaunch in early 2022, now integrating the new ICON-D2-EPS NWP data, and is being visited a few thousand times per day during flooding events. The approach to make relevant information public, also displaying data in form of maps and time series has proven to be successful. This goes in line with a movement towards open data, a strategy that we strongly support.
The warnings for three relevant water level gauges are also transmitted to the direct public warning apps NINA (https://www.bbk.bund.de/DE/Warnung-Vorsorge/Warn-App-NINA/warn-app-nina_node.html) and KATWARN (https://www.katwarn.de/) through the federal flood warning service LHP. Thus, every citizen in Hamburg with one of these two mobile phone applications receives a direct warning when a warning level is reached.
Warning systems building upon the experience from Hamburg have now been established for the city of Bremen (www.starkregenpartnerschaft.de), for parts of North Rhine-Westphalia (without public website yet) and are under construction for the cities of Flensburg and Cologne.