Thesis: Structuring the hydrodynamics of the Elbe estuary
- Partners
- University of Twente
- Researcher
- Karla Korporaal
Estuaries are the dynamic crossroads between rivers and seas, supporting vital ecosystems, commercial shipping routes, and flood protection. With the ever-increasing impacts of climate change, accurately modeling estuarine hydrodynamics has become crucial to understanding and managing these complex environments. Numerical models have emerged as indispensable tools for simulating estuarine processes, providing insights into tidal flows, storm surge impacts, and the effects of human interventions on water levels. Among these models, the subgrid method has shown potential for enhancing computational efficiency while capturing the intricate balance between wet and dry regions of estuaries.
This thesis research executed by Karla Korporaal explores the potential of combining a structured grid with subgrid modeling to improve accuracy and efficiency in the simulation of estuarine hydrodynamics, specifically applied to the Elbe estuary. By integrating a spatially and temporally varying tidal boundary and refining grid configurations, this research sheds light on the delicate balance between model accuracy and computational demands. The findings not only reveal critical insights into tidal dynamics but also offer practical recommendations for future estuarine modeling, making it possible to simulate complex estuarine environments more realistically in a fraction of the time. As we move forward, innovative modeling approaches like this are essential for building resilient coastal systems, ensuring safe navigation, and protecting estuarine ecosystems. This research, which achieved remarkable accuracy in simulating the Elbe estuary’s hydrodynamics, marks a step forward in our capacity to understand and protect these invaluable natural resources.
The challenge
Estuaries serve crucial roles, such as acting as navigation routes that link the sea to harbours.
Understanding the complex hydrodynamics of estuaries is essential for managing these vital
systems, especially when planning interventions or assessing risks. Numerical models are
useful tools that simulate estuarine hydrodynamics, offering insights into processes like tidal
movements and water level changes. They can predict critical conditions, such as water levels
during storm surges, essential for flood risk assessment (Boehlich & Strotmann, 2008). Various
numerical models exist which can simulate estuarine hydrodynamics, each with settings that
influence their accuracy and computational efficiency (Bomers et al., 2019; Bounagui et al.,
2003).
One distinction among numerical models is whether they incorporate the subgrid method. This
method offers significant advantages in estuarine modelling, including enhanced computational
efficiency and the ability to model partly dry/wet computational cells (Casulli, 2009; Volp
et al., 2013). Research on the Elbe estuary, Germany, has shown that unstructured grids with
subgrids improve accuracy and efficiency, but combining a structured grid with a subgrid has
not been explored thoroughly.
Our solution
This study investigates whether a model with a structured
grid and subgrid can accurately and efficiently capture estuarine hydrodynamics, and how
factors like tidal boundary, grid width, time step, grid refinements, roughness values, and
river discharge impact simulation results. This research uses the 3Di software, which supports
structured grids with subgrids and continues to use the Elbe estuary as study area.
The outcome
Instead of using a non-spatially and temporally varying tidal boundary, a more realistic
spatially and varying tidal boundary, consisting of multiple 1D boundaries, was implemented.
This boundary better captured small-scale water level variations in the sea region, while its
accuracy in the upstream regions was similar to the simulation with a non-spatially varying
tidal boundary. The settings of the 1D boundaries affected the total in- and outflow volumes
and momentum, influencing hydrodynamics across the estuary in both timing and magnitude.
To enhance natural flow patterns and prevent unrealistic high velocities and water levels an
extra line of computational cells was added at the boundary.
Next, an analysis on grid width and time step was conducted. Increasing the grid width reduced
tidal ranges in the river due to increased numerical diffusion, while larger grid widths resulted
in unexpectedly larger tidal ranges in the sea area. Further research is needed to understand
this effect. Larger time steps increased tidal ranges and delays, as they fail to capture certain widths or time steps reduce computational time but affect accuracy negatively.
Instead of using an uniform grid width, applying grid refinements can enhance both model
accuracy and computational efficiency. Strategically placing refinements in complex flow areas,
such as the river widening and main navigation channel, can improve the hydrodynamic results.
A coarse grid can be used in open sea regions if critical areas are properly refined. While
refinements enhance accuracy in specific regions, they can also increase computational time if
the number of computational cells increases. Thus, balancing grid width and refinements is
essential. For the Elbe estuary, refining the navigation channel is crucial.
Finally, the bottom roughness and river discharge were systematically varied to assess their
effects on hydrodynamic results. Lower roughness values led to larger tidal ranges due to
reduced energy dissipation. Additionally, river boundary influence was greater during low
tide, when tidal forces are weaker. Increased river discharge raised water levels, but its effect
decreased with distance from the river boundary.
In conclusion, this research show that model settings and schematizations significantly affect
simulated estuarine hydrodynamics and computational efficiency. The 3Di subgrid model shows
promise for large-scale estuarine studies, as it accurately captured the hydrodynamics of the
Elbe estuary with discrepancies of less than 10 cm compared to measurements, while requiring
just 5 minutes of computational time for a 15-day simulation.