Simulating wave setup and runup during storm conditions on a complex barred beach

The purpose of this study is to assess the ability of the SWASH model to reproduce wave setup and runup in highly dissipative stormy conditions. To proceed we use data collected during the ECORS Truc Vert'08 Experiment, especially during the Johanna storm in the winter of 2007-008 (wave setup under Hs= 8.2 m and Tp= 18.3 s and runup under 6.4 m and peak period up to 16.4 s). We test different model settings (1D and 2D mode) and model forcing (spectral and parametric) to reproduce sensor measured wave setup at several locations in the nearshore area and video measured runup on two beach profiles. For the whole tested configurations, the wave setup is reproduced accurately. Results considering all the sensor locations in the near shore area in 1D and 2D are significantly correlated to the observations with respectively ρ²=0.66 and 0.81; RMSE=0.13 m and 0.08 m without any significant bias. Observations and simulations of runup are investigated in terms of spectra and statistic component. 1D simulations produces an overall overestimation and no significant improvement is obtained by modifying the breaking parameters. The results for 2D simulations are fairly satisfactory reproducing significant swash height (S), but are significantly improved with spectral forcing than parametric with respectively ρ²=0.73 and 0.71, RMSE=0.19 m and 0.43 m. Generally, the model reproduces accurately the infragravity component but tends to overestimate the incident component, leading to an overestimation of the energy density for moderate wave conditions and more accurate results for higher-energy wave conditions. Results in 2D with spectral forcing show a saturation of the infragravity component with a threshold around Hs=4 to 5 m, which is comparable to the observations collected at Truc Vert Beach. As regards the conventional statistical parameter for runup estimation (R2%) three methods are applied to derive the 2% exceedence value for runup from observed and simulated shoreline vertical elevation time series. When R2% is based on the sum of wave setup and half of the significant swash height, results provided by the model are close or even better than estimations provided by empirical formulas from the bibliography. Defining R2% as the exceeded 2% values of the time, derived considering the cumulative distribution function of the entire water-level time series also provide fairly good results. Results using only runup maxima time series are less satisfactory. In the two last cases, R2% is slightly underestimated for moderate wave conditions (Hs<4 m; Tp ≈ 14 s) and overestimated for higher-energy wave conditions. Generally results shows that where extreme wave conditions are concerned, the model setting must be considered carefully because the simplification of 1D (rather than 2D), or the use of parametric wave description (rather than spectral), can be a source of significant inaccuracy or overestimation in simulated run-up values.