Improving XBeach non-hydrostatic model predictions of the swash morphodynamics of intermediate-reflective beaches

Abstract

A very common observation is the episodic erosion of beaches during storms and the slow recovery (accretion)
afterwards (Yates et al. 2009). Morphodynamic models parameterize physical processes in order to
relate the fluid motions (hydrodynamics) to the bed level changes (morphodynamics) over a wide range of
spatial and temporal scales. Despite recovery of the beach profile being a slow process, accretion mechanisms
in the swash zone are complex to represent by a numerical model due to the shallow, rapidly varying
flows and high concentration gradients (Brocchini & Baldock 2008). The swash zone on most beaches is
readily accessible, but this accessibility does not translate into a broad knowledge of the underlying physical
processes (Chardón-Maldonado et al. 2015). This research focused on assessing the representation of physical
processes in the swash zone of intermediate-reflective beaches during erosive and accretive conditions in
the non-hydrostatic version of the XBeach model (XBeach). This is a depth-averaged phase resolving model,
mainly used for calculations of storm impact (hours-days) on sandy and gravel beaches. Based on conclusions
in literature, relevant physical processes that contribute to accretion were determined. Using a simple
planar beach bathymetry, the sediment transport formulations in XBeach and the individual influence of
groundwater effects, bed slope effects, sediment response time and wave breaking induced turbulence were
assessed. The results showed that for both accretive and erosive wave conditions, XBeach predicts erosion in
the swash zone. Groundwater infiltration and wave breaking induced turbulence are likely to enhance onshore
transport significantly Reniers et al. (2013), Turner &Masselink (1998).
To verify the findings of the planar beach modeling approach, in the second part of this thesis the morphodynamical
predictions of XBeach were compared to the dataset collected during the Bardex II experiment.
Bardex II was performed in 2012 in the Delta Flume in the Netherlands and the dataset contains observations
of a series of experiments focusing on the effect of varying wave, sea level and beach groundwater conditions
on a sandy beach (D50=0.42 mm)(Masselink et al. 2013). Stored data and published resultswere used for comparison
with the morphodynamical prediction performance of XBeach. Two timescales have been analyzed,
the total morphological response of the beach on a timescale of 100 minutes and the intra-swash sediment
transport processes on a timescale of 10 seconds. Two experiments from the series have been reproduced.
The first, experiment A4, has an almost stable, slightly erosivemorphological response throughout the swash
zone. The second, experiment A8, shows accretion in the upper swash and a stable profile in the rest of the
swash zone. XBeach erroneously (over)predicted erosion above the mean sea level (MSL) for both A4 and A8
conditions. This conclusion was related to the results of the intra-swash sediment transport assessment. The
modeled velocity in both uprush and backwash were higher than the Bardex II measured velocity and XBeach
extremely underpredicted uprush sediment concentrations suggesting that turbulence induced by the bore
is not enough taken into account. Over-predicted backwash sediment concentrations for both accretive and
erosive conditions suggested that groundwater infiltration was not strong enough. However, enhancing wave
breaking induced turbulence and groundwater infiltration did not lead to an improvement of the predictions
of sediment concentrations in the swash.
The sediment transport formulations of XBeachwere developed using a long wave resolving (short wave averaging)
model, andwidely validated and calibrated on field and experimental data (Soulsby 1997, van Rijn et al.
2007, Van Thiel De Vries 2009). Therefore, in the last part of this thesis two possible improvements of the two
sediment formulations of XBeach (van Thiel-van Rijn and Soulsby-van Rijn) when applying them in a short
wave resolving model are assessed using a 1D sediment transport model. The decomposition of the velocity
signal in amean and a fluctuating part improved mainly the predictions using Soulsby-van Rijn where the
separate calibration of turbulent kinetic energy resulted in better predictions for both transport formulations.
This analysis was performed only at one point on the cross-shore domain (slightly above MSL). Although the
results for this position were promising, comparison of modeled sediment transport with Bardex II observations
at different positions throughout the upper and lower swash zone is needed to give a full validation of
the proposed adaptations to the transport formulations.