%0 journal article
%@ 1616-7341
%A Behrens, A.

%D 2015
%J Ocean Dynamics
%N 4
%P 469-486 
%R doi:10.1007/s10236-015-0825-y
%T Development of an ensemble prediction system for ocean surface waves in a coastal area
%U https://doi.org/10.1007/s10236-015-0825-y
4 
%X An ensemble prediction system for ocean surface waves has been developed and applied on a local scale to the German Bight and the western Baltic Sea. U10-wind fields generated by the COSMO-DE-EPS upstream forecast chain of the German Met Service (DWD: Deutscher Wetterdienst) have been used as the driving force for the third-generation spectral wave model WAM. The atmospheric chain includes four different global models that provide boundary values for four regional COSMO-EU realisations. Each of those drive five COSMO-DE members, respectively, with different sets of physical parameterisations, so that finally 20 members are available to run 20 corresponding wave ensemble members of the coastal wave model CWAM (Coastal WAve Model) for the German Bight and the western Baltic Sea. It is the first time that in an ensemble prediction system for ocean waves, an atmospheric model of such a fine spatial resolution of 2.8 km has been combined with a wave model running on a model grid with a mesh size of 900 m only. Test runs with the wave ensemble prediction system have been executed for two entire months (April 2013 and June 2014) and for an 8-day storm case (Xaver) in December 2013 in order to check whether such a system could be a reasonable step to improve the future operational wave forecasts of the DWD. The results computed by the different wave model members agree fairly well with available buoy data. The differences between the results for the integrated wave parameters of the individual members are small only, but more pronounced in extreme storm situations. Finally, the statistical analysis of the comparisons with measurements show without exception slightly improved values for the ensemble mean of the wave ensemble members compared with the usual deterministic routine control run.