Investigation of Wave Growth and Decay in the SWAN Model: Three Regional-Scale Applications

Abstract

Wave growth and decay characteristics in a typical wave action model [Simulating Waves Nearshore (SWAN)] are investigated in this paper. This study is motivated by generally poor agreement between model results and measurements for a regional-scale model of a two-day period during the SandyDuck '97 experiment, wherein there is consistent underprediction of lower-frequency (0.05?0.19 Hz) energy. Two separate methods are presented for improving predictions of low-frequency energy: 1) by altering the weighting of the relative wavenumber term that exists in the whitecapping formulation and 2) by disallowing the breaking of swell. The SandyDuck '97 simulation is repeated with the proposed modifications. Using the first modification, a slight improvement is seen, and with the second modification an apparent problem with nonphysical dissipation of swell by the model is corrected. The modifications are then applied to two other test cases, one in Lake Michigan and the other in the Mississippi Bight. Both cases are of similar scale to the SandyDuck '97 experiment but are freer of uncertainties related to forcing. In both cases, the underprediction of low-frequency energy is observed using the original model, and in both cases agreement with observations is improved via the first of these two proposed modifications. During the course of this investigation, it becomes apparent that, though the model's dissipation term can be improved by these modifications, fundamental problems with the form of the term severely limit the level of improvement that can be achieved.