Abstract
Three widely used wave models, namely, the open ocean wave model (Cycle-4.5, hereinafter referred to as WAM4.5) and the coastal models, Simulation of WAves Nearshore (Cycle III version 40.31, hereinafter referred to as SWAN) and the K-model, are applied to Lake Erie to simulate waves at a spatial resolution of about 4 km. The results of a three-week hindcast study are compared with buoy observations in terms of integrated parameters,
one-dimensional (1-D) and two-dimensional (2-D) energy spectra, scatter plots and statistical analyses
of the wave fields. The time development of the 1-D spectra by the models matches the buoy measurements well.
All the wave models tend to overpredict the wave heights and underpredict (particularly the K-model) the peak period. SWAN performs best for the wave heights and WAM4.5 for the peak periods and is computationally less demanding, whereas the spatial resolution applied to Lake Erie seems to be too coarse for an adequate use of the K-model. In general, WAM4.5 has advantages over coastal wave models in operational intermediate-scale applications.