Comparison between third and second generation ocean wave models

Abstract

The objective of this study is to assess the relative performance of a second generation (2G) model, WISWAVE (Hubertz 1992), and a third generation (3G) model, WAM (WAMDI 1988), in deep water. High-quality wind fields, with detailed spatial and temporal resolution, were used to force both models in an effort to isolate differences between their formulations. Even though previous studies (Cardone et al. 1996) have compared these models, this study is the first direct comparison where frequencies and directions have been discretized identically, no wind field interpolation was used and wave model grids are not nested. Five extra-tropical storm events which occurred along the US Atlantic coast were selected for comparison. All five historical wind fields were developed using kinematic analysis and objective techniques (Cardone 1992). These storm events were selected because of the availability of high-quality winds and their variety of conditions ranging from extreme events to more moderate and variable events characterized by sea and swell, shifting winds, and passage of cold fronts. Three of the storms occurred during the Surface Wave Dynamics Experiment (SWADE) (Weller et al. 1991) intensive operation periods (IOP) and are 1OP1, 1OP2, and 1OP3. The remaining two storms, the "Halloween Storm" and the "Storm of the Century", were significant events causing flooding and damage along the Atlantic seaboard. Each model was driven by identical wind fields, gridded to 1/2' latitude and longitude spatial resolution and 1-hour temporal resolution. Both models were run in the deepwater mode. Model results for these storms were compared to National Data Buoy Center (NDBC) buoy wave measurements. The analysis of the results is divided into three steps. First, a qualitative comparison is made between model predictions and buoy data for the mean parameters: significant wave height, mean wave period, peak wave period, peak wave direction, directional spreading coefficient, wind speed, and wind direction. Second, a series of statistical tests of the wave parameters was developed, including mean, bias, and RMS Error. In addition to the standard statistical parameters, a pattern correlation between model and measured 2-D spectra was presented.