There is a relative lack of research, targeted models and tools to manage beaches in estuaries and bays (BEBs). Many estuaries and bays have been highly modified and urbanised, for example port developments and coastal revetments. This paper outlines the complications and opportunities for conserving and managing BEBs in modified estuaries. To do this, we focus on eight diverse case studies from North and South America, Asia, Europe, Africa and Australia combined with the broader global literature. Our key findings are as follows: (1) BEBs are diverse and exist under a great variety of tide and wave conditions that differentiate them from open-coast beaches; (2) BEBs often lack statutory protection and many have already been sacrificed to development; (3) BEBs lack specific management tools and are often managed using tools developed for open-coast beaches; and (4) BEBs have the potential to become important in nature-based management solutions. We set the future research agenda for BEBs, which should include broadening research to include greater diversity of BEBs than in the past, standardising monitoring techniques, including the development of global databases using citizen science and developing specific management tools for BEBs. We must recognise BEBs as unique coastal features and develop the required fundamental knowledge and tools to effectively manage them, so they can continue providing their unique ecosystem services.

@en

Waves in coastal waters play a role in forecasting storm conditions, as boundary conditions for the assessment of coastal risk and in coastal structure design. Assessment of their impact can be made through numerical modeling. In a forecasting context, the CPU time of the wave model is decisive in its suitability. Recent years wave model approaches have been to describe both linear and non-linear generation, dissipation and transfer processes in the wave spectrum explicitly. This resulted in computationally demanding tools. Such models are very suitable for research purposes or for design and assessment if time is available, but are rather heavy tools for forecasting applications on larger scales. Parametric models describe the wave spectrum through characteristics in a smaller solution space compared to spectral models, which makes them potentially faster solvers. In this thesis the suitability for forecasting of a spectrally integrated approach to the wave action balance is investigated. The stationary wave drivers from the storm impact assessment model XBeach were taken as a starting point. A second balance equation was identified to model the evolution of the mean period. In estuaries and behind shoals, locally generated waves can be dominant over offshore wave conditions. As part of the project, a wave generation mechanism is developed for the modified wave drivers based on prediction relations for growth of wind-waves under fetch-limited conditions. The two-equation approach was discretized and tested in a 1D Matlab model and the solution was proven to converge towards the literature prescribed growth curves for total wave energy and mean wave period. This approach was then implemented in the unstructured Frequency Integrated wave drivers in D-Flow FM. The implementation has been verified through semi-1D test cases and showed convergence behavior equal to the Matlab model. A hindcast of a 1982 storm in the Haringvliet was performed to compare the model skill of the unstructured frequency integrated model (FIM-FM) to the spectral wave model SWAN. In this hindcast the model skill of FIM-FM is comparable to SWAN for significant wave height and mean wave period for all locations apart from directly behind a shoal. Depth and wave height dependence of bottom friction coefficients should be studied in more validation cases to improve the prediction of wave steepening over shoals. Further research should also look into the iterative method in the solver for stationary solutions of the model equations to realize faster CPU time of a stationary problem. Although not established in this thesis, faster CPU time compared to a spectral model like SWAN should be possible based on the reduction of the solution space.