Coastal response during the 1953 and 1976 storm surges in the Netherlands. Field data validation of the XBeach model

Abstract

The storm surge early-warning system that is going to be established in the Netherlands, combines the accurate weather forecast with the hydrodynamic and morphodynamic models in an operational mode, in order to estimate the potential impact on the coasts. This study focuses on the morphodynamic validity of the operational model system, by studying two historical storm surge events on prototype scale. Due to the lack of warning, the 1953 storm surge left behind thousands of casualties and extensive wreckage of the Dutch coastline. In order to reduce the probability of experiencing again such a devastating storm surge, the coastal defense policy in the Netherlands had been reorganized on national level and more effective countermeasures had been received. When the 1976 storm surge attacked the country, the civil awareness and the reinforced coastal defense abate the impact and the fatalities. The degradation of the coastline has been recorded as part of the Jarkus coastline monitoring programme, and after the conducted analysis the non-uniform impact along the North Holland province became evident. Two numerical models form the operational model schematization; the deformation and the propagation of the storm surge are simulated with the Delft3D model and the nearshore hydrodynamic and morphodynamic processes by the XBeach model. The available storm surge level records attest that the model underestimates the storm surge level for both of the events, up to 0.70m and the phase lag up to 45 minutes. In the studied site of Bergen, the performed sensitivity analysis of the XBeach model proves that the water level underestimation is an important parameter for the deviations between the coastal profile records as a result of the 1976 storm surge and the computed post storm profile, as higher computational skill is obtained when imposing the measured storm surge level record instead of the computed one. Furthermore, the XBeach model reacts as expected at the changes concerning the waves' asymmetry and the slumping of the water area through its avalanching mechanism, reflecting a higher computational skill. In contrast, the model is insensitive to the long waves' sediment stirring and to the additional imposed wind setup. Concerning the model performance in the area of Castricum, the model skill is excellent and very good convergence is obtained on estimating the volume change and the estimated dune retreat. In Julianadorp, while the influence of the groins is not accounted, due to the significant scouring that is observed in the backshore zone, the model performance is bad. The geological features and the bed profile of the studied areas present a different pattern of the energy distribution, which may indicate a possible reason for the longshore erosion variability during the 1976 storm surge. The operational model needs to be further verified with the 2DH simulations in order to account for the longshore profile development and the effects of the hard non-erodible structures, while it is also recommended to further investigate the extensive scour development in the backshore zone.