Shaken or stirred?

Abstract

Where rivers meet the sea, deltas form: branched, dynamic coastal ecosystems that have historically been attractive places for human settlement. Within these deltas, fresh riverine water and saline water meet in transitional zones, estuaries, and at the point where the river flows out into the sea, a river plume may be formed. To provide humans with services such as food, drinking water, and shelter, many deltas have been heavily modified. At the same time, these deltaic systems are typically highly vulnerable to such environmental change, so that the combination of human pressures and climate change can significantly amplify risks related to freshwater resources, ecosystem biodiversity, and coastal safety.

While climate-change-intensified flood risk has been extensively studied, risks to freshwater availability, such as salt intrusion, have received less attention. Salt intrusion is controlled by a complex interplay of river discharge, tides, wind, stratification, and human interventions. The combined effect of these processes on salt intrusion remains poorly understood, particularly in complex, multi-branch deltas or estuarine networks, such as the Rhine-Meuse Delta.

The overarching objective of this thesis is to gain more insight into salt intrusion dynamics in the coast-delta system of the Rhine-Meuse Delta. Understanding these dynamics is critical for both scientific and management purposes. This dissertation takes a modelling approach to addressing this objective. However, currently available realistic hydrodynamic models of the region often do not resolve all relevant scales, and analysis tools are unable to extract the underlying processes from the complex datasets they generate. Thus, this work is structured into two complementary parts: the first focuses on developing a suitable model and analysis tools, and the second applies these tools to investigate salt intrusion dynamics.

First, the scientific basis of this dissertation is established by reviewing the physical mechanisms governing salt intrusion in estuaries. The multi-scale nature of these processes is highlighted, from small-scale mixing to estuary-scale exchange flows. For salt intrusion in the Rhine-Meuse Delta, the most important processes are shown to occur on the scale of meters to the entire coast-delta system and on the intratidal to fortnightly timescale. Identifying these scales helps guide the choice of analysis methods. To extract the underlying processes from complex, realistic models spanning all these scales, this thesis analyses terms in the salinity variance analysis.

Identifying relevant scales for salt intrusion also informs hydrodynamic model selection. For this thesis, the realistic, three-dimensional, unstructured Delft3D Flexible Mesh Rhine-Meuse Delta model is chosen. The model is further developed to increase its predictive capability for an average-discharge year by adding a heat flux model and improving some of the boundary conditions. This new version of the RMD model is then validated for an average-discharge year using an extensive set of measurements in the region. This validation shows excellent reproduction of water levels, temperature, and good reproduction of salinity throughout the estuarine domain, thus ensuring that this model provides a reliable tool for investigating the mechanisms that control salt intrusion in the Rhine-Meuse Delta.

Next, the challenge of applying the chosen analysis methods to the RMD model is considered. Models with staggered unstructured grids, like the RMD model, are made up of polygons that can have any shape, as long as they adhere to certain orthogonality constraints. This, combined with the way datasets of models with staggered unstructured grids are structured, results in large, complex model output datasets. To circumvent computationally expensive calculations on these types of grids, output is often interpolated onto structured grids. However, the salinity variance analysis requires volume conservation and can thus not be applied to interpolated data. It is found that three practical implementation strategies are key to enabling complex computations on larger-than-memory unstructured datasets: unchunking alignment dimensions, ensuring perfect chunk alignment for element-wise operations, and strategically circumventing automatic alignment operations. A scientific software Python package, pySVA, is developed to apply the novel practical implementation strategy to model output datasets generated by the RMD model.

Using the tools developed in the first part of this thesis, the feedback between wind direction, river plume distribution, estuarine exchange flow, and salt intrusion is explored. Although river plumes are formed at the outflow of estuaries and the two are thus logically connected, it is unclear whether there is a dynamic two-way coupling between the estuary and the river plume — i.e., whether the river plume also influences the estuary. This is investigated by considering the influence of wind on the Rhine-Meuse Delta through the modification of the Rhine river plume. It is found that wind (direction and speed) modifies the plume's extent, stratification, and the strength and location of vertical mixing. Upwelling winds stretch and thin the plume, strengthen stratification, and influence the estuary by enhancing exchange flow through a shoreward bottom current. Downwelling winds thicken and narrow the plume, oppose estuarine circulation via altered cross-shore residual flow, and reduce exchange flow strength. Furthermore, downwelling winds can completely suppress cross-shore straining by prolonging the alongshore flood. Onshore winds attach the plume to the coast and suppress cross-shore straining, but through a different mechanism than under downwelling winds: by directly altering the cross-shore flow. Onshore winds decrease the cross-shore residual flow but increase the depth-averaged onshore current, resulting in relatively large exchange flows but low salt exchange efficiency. Thus, plume changes propagate into the estuary, altering the stratification, residual velocity profiles, and, ultimately, the strength and composition of the estuarine exchange flow. The net effect is that upwelling winds generally reduce salt intrusion, while downwelling and onshore winds increase it.

Finally, the effects of human intervention into the Rhine-Meuse Delta system are explored by considering the scenario of adding a second sea connection to the Rhine-Meuse Delta through permanently opening the Haringvliet floodgates. To assess this, two simulations with the RMD model are conducted: a reference scenario representing the average-discharge year and a scenario with permanently open floodgates and the same forcing as in the reference. It is found that the intervention changes the discharge distribution, alters tidal wave propagation and phasing, and reduces stratification in the northern branches of the estuary and the river plume. Placing the branches into the estuarine parameter space (Geyer & MacCready, 2014) shows that, as a result, almost all branches in the estuarine network undergo regime changes, and the salt intrusion length increases in all branches — an effect that cannot be explained by the commonly used steady-state salt budget. Additionally, comparing the two scenarios provides insight into the processes affecting salt intrusion in the Rhine-Meuse Delta in more general terms. Counterintuitively, analysis of the salinity variance analysis reveals that positive straining is negatively correlated to salt intrusion length, which challenges our current understanding of salt intrusion in salt wedge estuaries. The key to understanding this is to realise the importance of horizontal dissipation. Horizontal dissipation is found to be strongly correlated with straining, which is hypothesised to be related to baroclinic convergence and frontogenesis. Additionally, horizontal dissipation and the salt intrusion length are strongly negatively correlated, suggesting that horizontal dissipation is an important factor affecting salt intrusion in the Rhine-Meuse Delta.

Overall, this thesis demonstrates how coupled estuary–plume dynamics, advanced hydrodynamic modelling, and process-based analysis methods can be integrated to quantify and understand the physical drivers of salt intrusion, offering new insights into estuarine behaviour and informing sustainable management of complex deltaic systems.