DWD 2nd SRNWP Workshop on Short Range Ensemble, Bologna, 7-8 April 2005 Deutscher Wetterdienst Zentrale Vorhersage 1 • EPS: huge amount of information - NWP model pool - ECMWF EPS product family - COSMO-LEPS, SRNWP PEPS - previous EPS (or model) runs • decision for a certain szenario in a limited time fra 0. Introduction • Severe weather prediction by ensemble prediction tool it´s a challenge - great variability of weather even • Forecaster has to deal with the predictability of the the cost-lost-ratio, customer needs, model physics . 1. First steps - and still favorable used at the • AFREG-MIX: developed during the eighties in Potsd • low cost system (CPU-time) • Mixing GME and ECMWF and deriving weather parame by statistical methods - Metgrams, tables 2. Using an EPS - from the forecasters point of - automatically generate fc´s (AFREG, GMOS) - forecasters experience Questions: 1. Is it possible to add value to forecasts by simple m of NWP models without of any calibration ? 2. The main szenario provided by the EPS or the majorit the NWP models - is this always the correct one ? 3. Could a NWP model PEPS outperform the ECMWF EPS ?