Implementation and evaluation of coupled discontinuous Galerkin methods for simulating overtopping of flood defenses by storm waves

•“Loosely” coupled “operational” DG numerical shallow-water and wind-sea models.•Compute wave-overtopping flow rates at barriers by EurOtop guidance.•Numerical “wave flume” experiments show good agreement to physical measurements. In coastal regions, accurate prediction of the inundation risks posed by tropical cyclones and other severe weather is key to mitigating disaster impacts. Predictions of these risks are often provided by numerical models; however, these models also frequently do not consider inundation due to wind-sea wave run-up. Neglecting such effects may be misleading for evaluating the risks posed to areas protected by flood defense systems such as levees and dykes. To effectively evaluate the flooding risks posed by storm waves, accurate wind-sea modeling should be undertaken in parallel with storm surge modeling, and accurate wave-overtopping flow rates at flood barriers must be prescribed based on the modeled sea conditions and structure geometries. To this end, we present development of a loosely coupled finite element based shallow-water equations and parametric spectral wave model. Wave-overtopping flow rates are computed by formulas provided in the guidance of the EurOtop manual. We demonstrate that the parametric spectral wave modeling approach is sufficiently accurate for modeling of wave-overtopping events, despite its simplifying assumptions, such as those based on deep water (e.g., discounting foreshore effects such as wave shoaling and breaking), and disregarding internal forces due to wave breaking and turbulence, as opposed to more commonly used discrete (third generation) spectral wave models, which provide more detailed descriptions of wave fields but are much more computationally expensive. Comparisons of numerically modeled wave-overtopping flow rates with measurements from the CLASH database show good agreement by linear regression (slope: 0.9911, R2: 0.9584) for more than 90% of the test cases considered, where for the relatively few outlier cases the model performs poorly due to either uncertainties in the EurOtop guidance or the lack of foreshore effects in the wave model.