Improvement of the numerical efficiency of SWAN: SBW Waddenzee

The spectral wind wave model SWAN (Booij et al. 1999) plays a key role in the estimation of the Hydraulic Boundary Conditions (HBC) for the primary sea defences of the Netherlands. Since some uncertainty remains with respect to the reliability of the wind wave model SWAN for application to the geographically complex area of the Wadden Sea, a number of activities have been initiated under the project ‘SBW Waddenzee’ to devise a strategy for the improvement of the model. In this regard, hindcast studies carried out with SWAN for the Amelander Zeegat in the Wadden Sea have shown that significant computational times are required to achieve results with the desired levels of numerical accuracy. This has prompted investigations into ways of reducing the computational time of SWAN. In the present study, various methods for reducing the simulation time of SWAN have been evaluated and refined, to facilitate the efficient execution of large numbers of SWAN simulations (possibly in combination with parallel computing) during the HBC computations. The methods evaluated here include the following: the Dynamical Deactivation Method (DDM), the Multigrid (MG) method, deactivating the action limiter and quadruplet interaction in the surf zone, improving the efficiency of the DIA quadruplet interaction term, and batch run organization using the so-called hotfile functionality of SWAN. These methods were all evaluated using the same set of eight field cases (featuring the Amelander Zeegat, the Eems-Dollard, Petten and Lake Sloten), and the same evaluation criteria, taking into account the total CPU time, iteration behaviour and mean convergence errors. This study has shown the DDM and batch run approaches to be the most successful techniques, both yielding reductions in total simulation time for Wadden Sea applications in the order of 31-35% with respect to the default SWAN model. The use of these methods affect model outcomes, which for the significant wave height and mean period can
SBW