One Two
Sie sind hier: Startseite EXTERNAL PARTNERS Bayesian Postprocessing of Ensemble Forecasts

Bayesian Postprocessing of Ensemble Forecasts

Deutscher Wetterdienst, Offenbach - Sabrina Bentzien, Petra Friederichs and Andreas Hense, Meteorological Insitute, Universität Bonn

Weather forecasts on the convective scale over periods of 12-24 hours represent medium range forecasts. Accordingly, forecast uncertainty is large. Probabilistic weather prediction accounts for the uncertainty and enables the estimation of probabilities for rare or extreme weather systems. This projects developes ensemble postprocessing methods with special emphasis on high-impact weather (extremes). We develope Bayesian methods for the interpretation and calibration of an ensemble prediction system (EPS) using the convection-resolving NWP model COSMO-DE.

Probabilistic forecasts enable the forecast users such as water management authorities to optimize action according to their individual cost-lost-ratio. The quatification of predictive skill gained by COSMO-DE-EPS in comparison to other reference approaches provides guidlines for issuing forecasts of high-impact mesoscale weather.


Bentzien, S. and P. Friederichs (2013): Decomposition and graphical portrayal of the quantile score. Q. J. R. Meteorol. Soc., doi: 10.1002/qj.2284

Bentzien, S., and P. Friederichs (2012): Generating and calibrating probabilistic quantitative precipitation forecasts from the high-resolution NWP model COSMO-DE. Wea. and Forecasting 27, 988-1002, doi: 10.1175/WAF-D-11-00101.1