André Walser MeteoSwiss, Zurich Evaluation of the COSMO-LEPS forecasts for the floods in Switzerland in August 2005 and further recent results… André Walser MeteoSwiss, Zurich
Case study: Flood event in Switzerland in August 2005 Photos:Tages-Anzeiger
Synoptic overview: 22 August 2005 Temperature 850 hPa and geopotential 500 hPa: 18º 10º 2º
Total precipitation over 3 days (20. – 22.8.) (06 - 06 UTC) C. Frei, MeteoSwiss About 400 stations, precipitation sum locally over 300 mm!
COSMO-LEPS forecast for 72h precipitation
COSMO-LEPS forecast for 72h precipitation
Probability precipitation > 100mm/72h C. Frei, MeteoSwiss
Probability precipitation > 100mm/72h C. Frei, MeteoSwiss
Probability precipitation > 250mm/72h C. Frei, MeteoSwiss
Using COSMO-LEPS for hydrologic forecasts Hydrologic model setup: Rhine upstream of Rheinfelden (34550 km2) Model: PREVAH (Gurtz et al. 1999) Driven by COSMO-LEPS A few test cases with 51 ensemble members Reuss (Luzern) Mark Verbunt (ETH Zurich) et al., submitted to J. Hydrometeorology
Case study: Ensemble runoff prediction for flood event 1999 Mark Verbunt et al., submitted to J. Hydrometeorology
Case study: Ensemble runoff prediction for flood event 1999 Mark Verbunt et al., submitted to J. Hydrometeorology
Synop-Verification: COSMO-LEPS & aLMo aLMo: - deterministic 7 km model (45 levels) of MeteoSwiss for 72h forecasts - same code (LM) as COSMO-LEPS aLMo domain Common verification domain, about 1000 synop stations
Standard deviation: 2 m temperature 12 UTC DJF 2004/2005 JJA 2005 Forecast range (h) Forecast range (h) uw: unweighted / w: weighted according to cluster population
Standard deviation: 2 m temperature 00 UTC DJF 2004/2005 JJA 2005 w Forecast range (h) Forecast range (h) uw: unweighted / w: weighted according to cluster population
Summary COSMO-LEPS forecasts for 2 case studies provides early warnings for extreme events. Coupled high-resolution atmospheric-hydrologic EPSs have a great potential to improve runoff forecasts. COSMO-LEPS ensemble mean for 2m temperature has better skill (lower standard deviation to observations) with weighting the ensemble members according to cluster populations compared to the mean with unweighted members.