Comparison stationary vs non-stationary SWAN runs : Wadden Sea hindcast 5/6 dec 2013

In order to assess the performance of SWAN in the different computational modes (stationary, quasi non-stationary and non-stationary) a hindeast study is carried out for the storm that occurred during the 5th and 6th of December 2013 in the tidal inlet of Ameland. The set of eight existing stationary simulations (Deltares, 2014a) has been extended with quasi- and non-stationary runs covering the period of 3 - 7 December 2013. Various non-stationary runs were done, with different time steps and maximum number of iterations. The results of the different computational modes are compared in terms of accuracy, iteration behaviour and computational time. In general, a stationary simulation is considered to be justified when the residence time of the simulated waves, i.e. the time that waves require to propagate through the model domain, or the time scale of local wave generation (in i.e. Lake IJssel or locally generated waves in the Wadden Sea) is small relative to the time scale of changes in the wave boundary conditions and forcings (e.g. currents, water levels and wind). For the concerned cases in the storm of 5/6 December 2013, the differences between the results with stationary SWAN and non-stationary SWAN are limited. Differences with the observations are in general larger. There is no general trend visible that stationary results with respect to significant wave height and wave period are always higher or lower than non-stationary results, nor for increasing conditions before the storm peak or decreasing conditions after the peak.