Morphodynamic modelling of an embayed Mediterranean beach: effect of the forcing sources

<p align="justify">Modelling the response of sandy beaches to sea level rise is a major scientific challenge and several types of models can be applied. Process-based models are useful tools for understanding beach responses on short time scales, but they are computationally expensive and tend to accumulate errors when resolving the many short-term processes they contain. Alternatively, reduced-complexity models can be an interesting option for long-term modelling. Furthermore, to simulate the long-term response of beaches, it is necessary to combine different model forcing sources (wave and sea-level, e.g., from buoys or hindcast models). The aim of this contribution is to quantify the effect of using different forcing sources in morphodynamic modelling with these model types.</p> <p align="justify">Two numerical models are applied to the embayed microtidal beach of El Castell (Palam&#243;s, Catalunya, NW Mediterranean). The XBeach process-based model, which solves the full 2DH nearshore hydrodynamics and the corresponding bed evolution, and the Q2Dmorfo reduced-complexity model, which simulates the bed level variations by calculating the sediment fluxes parametrically directly from the wave field without resolving the currents. Both models are first calibrated using two topobathymetric surveys conducted in January and July 2020 and wave and sea-level data measured from an AWAC deployed at 14.5 m depth during those 6 months. Calibration is performed using the most sensitive parameters, i.e., those related to cross-shore transport. In XBeach, the surfbeat mode must be used to obtain realistic results. It generates an aleatory spectral wave time series at the boundary that includes groupiness. To handle the effect of this randomness, a total of 5 realizations are made for each set of parameter values and a mean bathymetry is computed out of these realizations. To assess the model performance, the Brier Skill Score (BSS) is calculated both for the modelled bathymetry and its coastline during the 6-month period. Moreover, the Standard deviation (STD) of the 5 realizations is also computed. The chosen optimum parameter setting is the one maximizing the BSS (0.36 and 0.79 for the bathymetry and shoreline respectively) and minimizing the STD so that the result is robust and reproducible. In Q2Dmorfo, the BSS of the simulated final coastline is calculated for each set of parameter values. The optimum parameter setting also produces a maximum BSS of 0.79.</p> <p align="justify">Once both models are calibrated, the other potential forcing sources are applied. They include wave and sea-level datasets from a sea-level and wind-waves 72-years hindcast generated with the hydrodynamic-wave coupled SCHISM model, a wave dataset from an offshore buoy propagated to the AWAC position using SWAN model and a sea-level dataset measured by the tidal gauge at Barcelona harbour. The results show a very significant sensitivity to the wave forcing source and much less sensitivity to the sea-level source. By using the dataset propagated from the buoy by SWAN, both models represent well the observed beach rotation, whereas using the dataset obtained with SCHISM, the beach rotation is systematically under-predicted by both models, giving negative BSS values. This is because SCHISM predicts wave angles biased to the west.</p> <p>&#160;</p>