The Transnational Forum No. 4 was dedicated on «Quantitative Precipitation Forecast as Input for the Operational Flood Forecast». It took place in Trier, on March 4, 2008. About 40 specialists from different flood forecast centres in Germany, France and Luxembourg participated to this knowledge exchange.

Presentations

The following presentations were held:

•  Présentation générale du projet TIMIS (291.84 kB) (Ivo Leiss, Ernst Basler + Partner AG)
•  Grundlagen der numerischen Wettervorhersage unter besonderer Berücksichtigung der quantitativen Niederschlagsvorhersage (2.82 MB) (Clemens Simmer, University of Bonn)
•  La prévision numérique à Météo-France et les produits disponibles pour la prévision des crues (1.3 MB) (Jean-Michel Soubeyroux, Météo France)
•  Deterministische und probabilistische Niederschlagsvorhersagen beim Deutscher Wetterdienst (1.64 MB) (Detlev Majewski, Deutscher Wetterdienst)
•  Application des prévisions quantitatives de précipitation en modélisation hydrométéologique La chaîne SIM couplée à la prévision (1.85 MB) (Jean-Michel Soubeyroux, Météo France)
•  Die quantitative Niederschlagsvorhersagen der operationellen Wettervorhersage – Modelleinschätzung und Interpretation der Ergebnisse (2.32 MB) (Norbert Bonanati, Deutscher Wetterdienst)

Conclusions and Findings

The relevant issues, procedures and limitations were discussed. The following statements and conclusions have to be considered:

• There is a shift in responsibility from humans towards machines. Otherwise, the huge amount of data cannot be handled and analyzed.
• The human knowledge still remains important also when forecast systems were improved in the last years.
• The ensemble prediction systems show encouraging results, however they are also risky because of the huge amount of data to be processed and evaluated .
• The use of different models could also improve the reliability of forecast information. Due to the fact, that a catchment area often is dedicated to a specific hydrological model, this additional information is not taken into account.
• The meteorological network for snow melt is not dense enough in order to allow a 2-dimensional interpolation. Therefore, this parameter needs to be modelled using other parameters.
• The quantitative precipitation forecast is currently hampered by two problems: (1) The precipitation forecast is not reliable for a single geographic point. (2) The daily cycle of convective precipitation is difficult to forecast. Both issues are current research topics.
• The customers of the forecast information are used to single value forecasts . However, it is in the nature of things that forecast results are uncertain.
• For the customers, but also between the involved authorities, it is important to have one contact person.