@misc{74704,
  author = {Thomas Haiden and L. Magnusson and Ivan Tsonevsky and Fredrik Wetterhall and L. Alfieri and Florian Pappenberger and Patricia de Rosnay and J. Munoz-Sabater and Gianpaolo Balsamo and C. Albergel and Richard Forbes and Tim Hewson and S. Malardel and David Richardson},
  title = {ECMWF forecast performance during the June 2013 flood in Central Europe},
  abstract = {A case of orographically enhanced, extreme precipitation which led to extensive flooding in Central Europe is investigated. The ECMWF Integrated Forecasting System (IFS) provided good forecasts of the location of the heaviest precipitation on the northern side of the Alps but underestimated the magnitude. As a result, streamflow predictions for the Danube by the European Flood Alert System (EFAS) run with ECMWF input underestimated peak discharge values. We investigate possible causes for the underestimation such as model resolution, the microphysical parameterization, and soil moisture in the model. Increasing the model resolution has a substantial positive effect on the magnitude of the predicted rainfall in the worst affected region on the northern side of the Alps. It also improves forecasts over the Czech Republic where an observed, non-orographic band of enhanced precipitation is predicted which was not simulated in the operational forecast. A new parameterization of precipitation formation which will become operational in 2014 is shown to have a positive effect as well, leading to about 10% increase in total precipitation on the northern side of the Alps. Experiments with different options for soil moisture assimilation show little effect since the near-saturated conditions due to positive rainfall anomalies in the weeks prior to the event were already well captured by the soil model in the IFS, even without soil moisture assimilation.},
  year = {2014},
  journal = {ECMWF Technical Memoranda},
  number = {723},
  pages = {34},
  month = {06/2014},
  publisher = {ECMWF},
  url = {https://www.ecmwf.int/node/9731},
  doi = {10.21957/4p9ebc2r3},
  language = {eng},
}