Validation of the numerical wave models SWAN and HISWA at Norderney

Abstract

In coastal engineering waves often play a dominant role. To predict the wave behaviour in coastal regions numerical wave propagation models have been developed. This study deals with the verification of two models against measurements in the field, namely HISWA and SWAN. The second generation model HISWA has already proven to perform well for both engineering and research purposes. The third generation model SWAN is a new model, still under development. Also made by Delft University of Technology, SWAN is more or less the improved sequel to HISWA. Both models are based on the action density balance, which is made stationary to reduce the required computer capacity. They simulate wave growth and decay accounting for wind input, bottom friction, depth limited wave breaking, and whitecapping. The main differences between the models are the fact that SW AN is spectral both in frequencies and in directions, whilst in HISWA the frequencies are parametric. Secondly, the numerical scheme in SWAN for the wave propagation is unconditionally stable and encompasses waves from aU directions (360°). In HISWA the directional sector wherein wave propagation is considered is limited to a maximum of 1200 to obtain stability. Furthermore - being third generation - SWAN allows the spectrum to develop without any a priori constraints. During the measuring campaign, which was carried out by the Coastal Research Station Norderney, in the winter of 1995-1996, nine Waverider buoys recorded wave data. These buoys were located in the Norderneyer Seegat, a Wadden Sea area. Two cases have been simulated with SW AN and HISWA: one during high tide and one during low tide. I varied several parameters to examine the influence on the wave heights. These variations concern wind input formulation, wind speed, nonlinear wave wave interactions, friction, breaking, whitecapping, water level and incoming wave height. With reference to these research runs a set of input parameters is composed with which the models perform best in the area of this research. Both models tum out to have a good performance indicator (>90%). It has been found that the wave heights are to a large extent determined by the wind and by breaking. The wave height of the incoming wave at the seaward boundary of the computational domain is of minor importance. In equal circumstances, SWAN tends to calculate higher wave heights than HISWA. It appeared from this study that SWAN performs slightly better than HISWA and is therefore a suitable tool for wave prediction in a Wadden Sea area.