Wave attenuation due to vegetation: Duursche Waarden (case study)

Abstract

Conventional dike reinforcement measures do not take vegetation into account, while vegetation contains wave damping properties. Reasoning for this is that vegetation in general is seen as temporary and hence may be removed in the future. In this case study for the Dutch waterboard WDOD, the wave damping properties of vegetation in a floodplain along the river IJssel (Duursche Waarden) are studied. Duursche Waarden consists of mostly willow species and is adjacent to a dike. It is part of Natura2000 and hence it is a protected vegetation area and not considered to be temporal. A general approach is developed for similar sized (riparian) forests as Duursche Waarden (1.1 km2) to map the vegetation and obtain the vegetation parameters. This approach is also applied to Duursche Waarden. First of all the different vegetation areas within the riparian forest are identified with the help of aerial photos and field observations after which they are prioritised with the help of a Multi Criteria Analysis. The prioritised vegetation areas are divided into uniform and non-uniform areas and studied in detail. Uniform and non-uniform areas are distinguished based on tree species, structure, height and density with the help of field observations, aerial photos and airborne Light Detection and Ranging (LiDAR) data. A representative distribution of the vegetation parameters over the different vegetation areas is obtained with the help of one method. The vegetation parameters interesting for wave damping are the frontal area per volume and the bulk drag coefficient of leafless vegetation. Leafless vegetation is of interest since most extreme storms occur in winter. On top of that leaves seem to have insignificant contribution to wave damping in the full scale physical experiments of white willow trees conducted in the Delta Flume at Deltares. These experiments were performed in 2018. A new definition is introduced for the frontal area per volume. The frontal area density distribution over the height is estimated with the help of point clouds retrieved from TLS measurements and validated with hand measurements and studies of similar vegetation. Two methods are used in estimating the frontal area with the help of literature: the alpha shape and grid method. These methods are used for merged point clouds from Multiple Scanning Stations (MSS) and a point cloud from a Single Scanning Station (SSS). The SSS method includes a shadowing correction factor. Finally, the values of the frontal area per volume are chosen based on assessing the outcomes of the MSS method, SSS method and hand measurements. The bulk drag coefficient is estimated using selected values from literature. Dense vegetation contains relatively higher values for the vegetation parameters than sparse vegetation. This will result in more wave energy dissipation and hence more wave damping for dense vegetation. The frontal area per volume of sparse to moderate dense woody vegetation (values on average of about 0.01-0.20 m-1) in uniform areas is believed to estimated best and most efficiently with the MSS method. In non-uniform areas hand measurements are believed to be estimated best and most efficiently by hand measurements. The frontal area per volume of dense woody vegetation (values on average greater than 0.20 m-1) is most uncertain. Therefore, both hand and TLS measurements (with the SSS method) should be used to estimate the frontal area per volume. The wave attenuation by vegetation is estimated using the numerical wave model SWAN 1D and 2D computations including and excluding vegetation using the obtained vegetation parameters. No currents are included. Hydraulic conditions of Hydra-NL calculations using Bretschneider are used for both SWAN 1D and 2D computations. A comparison between Hydra-NL and SWAN 1D computations including model uncertainty for exactly the same conditions show similar outcomes for SWAN. SWAN 1D computations showed wave attenuation due to vegetation with a magnitude of 0 to about 44 cm for all dike locations adjacent to Duursche Waarden. SWAN 2D computations without vegetation show significantly lower outcomes than 1D computations without vegetation: on average about 12 cm. This difference is explained by refraction and directional spreading resulting from the spatial variability in water depth and fetch of the studied area. Therefore the studied area is believed to be better described by 2D computations. One combination of the wind direction and water level is proven to be governing for all dike locations. Following 2D computations, the failure mechanism ‘erosion of the outer slope’ with a return period of 66666 year results in an estimated maximum wave height of 77 cm for dike locations adjacent to Duursche Waarden. This is a reduction of about 23 cm compared to the results of Hydra-NL. Indicative calculations suggest that dike reinforcements are still needed at Duursche Waarden. The wave damping is estimated to be 16 to 24 cm at dike locations adjacent to relatively large vegetation areas of sparse to dense vegetation. Dike locations at which the significant wave height is the greatest are called critical dike locations. The wave damping for these critical dike points is estimated to vary between 4 to 7 cm. In a report by Deltares an overview is made of promising wave damping foreshores in the Netherlands. Rough estimations show possible wave damping at Duursche Waarden of about the same order of magnitude as found in this study. Also studies on similar vegetation show significant greater values for the wave damping, suggesting rather conservative than optimistic outcomes in this study.