Post-Assessment of ECMWF-mHM ensemble flood forecasting for 2021 summer flood in west Germany
Abstract
The highest death toll caused by river flooding in Germany since 1870s was recorded in summer 2021 [1]. The flood peak has been estimated between 975-1000 m3/sec, four times larger than the previous highest flood peak observed between 1947-2019 (236 m3/sec). In this presentation, the complexities associated with hydrologic model calibration will be discussed for simulating and capturing the flood peak of such a rare event (108-year return period). For this purpose, the mesoscale hydrologic model mHM [2] is calibrated for the Ahr basin using high-resolution quantitative precipitation estimation provided by the German weather service (DWD) [3-5] as input to generate hydrologic initial conditions and streamflow forecasts.
To investigate the hydrologic predictability of the event, a post-assessment of flood forecasting performance was conducted using HS2S syetem. 51 real-time atmospheric model ensemble 15-day forecast from ECMWF are post-processed and fed into mHM. Although the flood forecast ensembles are underestimated as a result of biases in atmospheric forecast, the result show that the probability of exceedance from the flood design (HQ100) is more than 20 percent. The forecast improvement by applying a bias-correction method which preserves trend will also be discussed. This work is considered as a basis to develop a near-real time experimental flood early warning system for German river basins.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.H35I1221N