Long term modelling of a mediterranean embayed beach: reduced-complexity model vs XBeach model

The response of sandy beaches to the expected sea level rise during the XXI century is a major scientific concern. Process-based 2DH morphodynamic models are unable of making projections to 2100 due to the high computational coast and to the accumulation of errors in resolving short-term processes. Reduced complexity models are then an alternative to make such projections. However, these models need to be validated and this can be performed with data and/or with process-based models. With the final aim of making long term projections of an embayed beach with the Q2Dmorfo reduced-complexity model, we do such validation with the XBeach process-based model.

Cala Castell is an embayed sandy beach at the Costa Brava (Catalonia, Spain) about 300 m wide, bounded by rocky headlands and facing to the South. Bathymetric surveys were conducted on 28 January and 8 July 2020. During this period, an AWAC was deployed in front of the beach, measuring mean sea level and wave height, period and direction. Both models were calibrated to reproduce the observed coastline behaviour and the best Brier skill score was high for both, BSS=0.79, with final shorelines being similar among them and to the observed one. However, since intermediate bathymetric surveys are not available, it is not possible to compare how models perform during particular wave events. To shed some light into this, a number of synthetic events are here investigated on a synthetic beach based on the geometry of Cala Castell. The wave conditions are selected so as to mimic typical wave conditions at the site, where the dominant wave direction is quite oblique, from the East. The model parameter values are the optimum ones after the calibration.

The cross-shore transport is parameterized in a completely different way in both models and it is hardly comparable but the longshore transport for oblique wave incidence should be consistent. We focus the comparison in the latter by using slightly different initial bathymetries, so that each model starts from its “equilibrium” bathymetry. For Q2Dmorfo, the bathymetry is constructed from a parabolic curve approximating the initial observed shoreline and the optimum equilibrium beach profile of the calibration. For XBeach, we first run the model during 10-30 days with constant shore-normal wave conditions over the initial Q2Dmorfo bathymetry to obtain an equilibrium configuration. Then, when applying oblique wave incidence conditions there is in both models a similar tendency to beach rotation, with shoreline retreat in the central part of the beach and progradation in the downwave part, although important quantitative differences may arise. The best comparison is done by analyzing the volume accumulated in each cross-shore profile per alongshore distance unit (m3/m) and it is found that the volume of sand eroded or accumulated in each profile for both models compare quite well. Therefore, on the basis that XBeach is close to reality when simulating a single event, this confirms the capability of Q2Dmorfo to describe longshore processes reasonably well even at the event time scale.