Building with Nature BwN Guideline Environments Project phases Governance BwN Knowledge base
BwN Building Blocks BwN Toolbox Pilots and cases BwN Knowledge

Log in

Along coasts of large Dutch freshwater lakes, dikes often border the water directly, with relatively steep slopes. Shallow zones and the gradual slope from land to water are lacking. Consequently, species that inhabit these zones are decreasing. In addition, constant lake-water levels cause erosion of shores. To dampen waves and recreate gradual land-water transitions brushwood mattresses were constructed in front of the dike. These mattresses might facilitate development of floating reed marsh in the shallow zone in front of a dike. This marsh reduces wave impact on the dike, enhances sedimentation and creates a clear shallow water zone with (submerged) vegetation. Thereby, the initial substrate of the mattress could be suitable for establishment of filter feeders, such as zebra mussels (Dreissena polymorpha) and for other species.

Building with Nature Design Traditional Design

The innovative application of braided brushwood mattresses aims to create floating foundations for emergence of reed vegetation. The floating mattresses locally reduce currents and waves thereby decreasing hydraulic loads on the dikes and creating valuable habitats above and below water. More benign conditions stimulate settling of suspended solids and promote the stabilisation of silty soils.


Traditionally dikes in Dutch freshwater lakes directly border the water. However, historically this zone used to be a gradual transition between land and water inhabiting freshwater marshes. 
To reduce wave impact on dikes rows of poles can be placed in front of the dike, or dikes can be designed to withstand wave impacts themselves.

    General Project Description



    Floating Reed Marsh


    The Netherlands, Houtribsluizen Markermeer side


    2009 - 2010


    Deltares, Waterdienst, van Schaik B.V.


    Permanently floating brushwood mattresses to allow development of reed vegetation, to reduce wave impact on the dike and to restore shallow freshwater habitats.


    Wave attenuation, sediment entrapment, habitat restoration, water purification, floating wetlands.

    Project Objective

    The Markermeer (Lake Marken) is the south western part of the IJsselmeer (Lake IJssel). Lake Ijssel is the remainder of the shallow sea arm Zuiderzee after construction of a closure dam (1932) and extensive land reclamations between 1930 and 1968. The Markermeer area was designated to be the last reclaimed area, but the plans were cancelled as the need for agricultural land decreased. By that time the dike had already been built. Since then, the lake has become ever more turbid, due to organic SPM (soluble microbial products) and resuspended old marine clay deposits. This has led to a severe degradation of the ecosystem. The project NMIJ ("Natuurlijker Markermeer en IJmeer") aims at the improvement of the ecological quality of the lake by stimulating experiments with innovative measures (van Geest et al. 2010). The improved ecological quality should be reached in three steps:

    1. reducing suspended matter concentrations,
    2. enhancing habitat diversity and dynamics and
    3. increasing connectivity.

    Project Solution

    The floating marsh was designed to serve all three steps in improving ecological quality, while additionally sustaining dike safety (van Geest et al. 2010). The marsh will function as a wave attenuator, and thus reduce wave impact on the dike. Underneath the floating marsh, low-energy zones will exist that allow suspended matter to settle. If applied in combination with suitable open water areas, the marsh can create a sheltered water zone with clear water which can be colonized by submerged vegetation. The marsh can also serve as a spawning, feeding and sheltering habitat for a variety of species and as an ecological connectivity zone for flora and fauna on shorelines. The mattresses were constructed from brushwood (braided willow osier) in which reed rhizomes were planted. The construction and durability of braided willow osier mattresses is historically very well established as they are used for stabilizing sediment at the base of dike constructions. Brushwood is locally available in large quantities and can be seen as a sustainable and local source of material for this type of solutions.

    Project Results

    This first-built floating marsh did not deliver sufficient long-term floating capability (van Geest et al 2010). The floating capacity of the growing Reed rhizomes (they have natural air chambers) did not compensate the gradual loss of floating capacity of the brushwood mattress. Five months after start the mattress began to sink. Reed growth from the rhizomes on the mattress was limited by lack of nutrients. Although this experiment may seem unsuccessful at first sight, it has been successful as application for underwater mattresses. The combination of cheap material, transportability over water, applicability in autumn and winter and an open and slowly sinking structure allowing bottom-dwelling organisms to escape turns out to be attractive for bottom mattresses.

    Spin-off and open end

    Consortium partners are working to get floating marsh mattresses internationally applied for combating wave intrusion into opened up marshes, by reducing wind fetch, and for closing shallow and small channels without sediment additions. However, at present, there is a need for an improved design. The challenge is to design mattresses of biodegradable materials that have sufficient buoyancy and that do not break on anchoring points, while having a good anchoring mechanism. Extruded recycled plastics and biodegradable plastics are being looked into for applicability, but have limitations as their production is generally not mainstreamed.

    Planning and Design


    Several ideas were launched to improve the ecological quality of the Markermeer including the construction of wetlands and the implementation of soft vegetated foreshores. Creating floating wetlands with braided brushwood mattresses was considered a quick, multifunctional and cheap option. From other studies (e.g Coops et al., 1996 and Lovstedt et al., 2010) it is known that vegetation in front of a dike is able to reduce wave impact and stabilize and raise the shore by sediment trapping. In freshwater environments, such as the IJsselmeer, this vegetation will most likely consist of reed and willows. However, many parts of the shore are too deep for the development of such a foreshore. The concept of a floating marsh may allow development of wave-attenuating vegetation without the need to raise the bed with imported sediment. The floating marsh concept is based on existing natural floating wetlands and islands in Louisiana and in the Danube delta. Additionally, floatlands with vegetation are broadly applied in urban environments, such as in canals in Amsterdam.


    To assess feasibility of the floating mattress pilot factors such as construction technique, material use, floatation, and functionality were addressed. Braiding mats of willow branches is an old traditional Dutch practice, thus, techniques for construction and transportation of braided brushwood mats are available. The mats are generally used as sinking units to stabilize beds under water in preparation of dike construction. Mats are known to remain buoyant for considerable time.

    After successful establishment of reed, sinking of the mat and submergence of reed could be beneficial for reed growth, as reed needs dry periods for germination and establishment, but prefers wet conditions once established. Creation of sheltered areas below and behind the mats might enhance settling of suspended matter. At the lower submerged side of the mats filter feeders might be able to attach, and hiding- and spawning places for fish are created.

    Sinking speeds were difficult to predict and depend amongst others on quality of material and climate conditions. If mattresses can be implemented with a sufficient floating capacity over a sufficiently long time, other possible functions of the floating marsh could be:

    • storage of CO2,
    • nutrient and contaminant uptake,
    • biomass production for energy or fodder production,
    • sediment trapping below the mattress,
    • stabilization of soft and muddy beds, and
    • adaptation to varying water levels.


    In the feasibility phase, a conceptual design was finalized of floating wetlands constructed of bundles of brushwood (‘wiepen’ in Dutch). The willow wood should be harvested at least one year before, to prevent re-growth of willow sprouts from the brushwood. A filling layer of dry reed material is created in between the bundles of brushwood for extra floating capacity and to prevent initial grazing on the reed roots and shoots by geese (van Geest et al. 2010). Reed growth is initiated by planting rhizomes of Phragmites (van Geest et al. 2010). Within the marshes, open water areas are designed to increase the length of edges (not covered) and to increase areas of quiet clear water, available for utilization by animals and vegetation. Furthermore, open areas will allow light and oxygen to reach the water underneath the mattress and prevent water quality deterioration.

    The floating marsh has to be strong enough to resist wave action. The strength is determined by the strength of the connections of the bundles. Although not in accordance with the 'cradle to cradle' principle, the connections between the bundles of brushwood are made of metal wire in order to guarantee sufficient strength. 
    The reed marsh was first tested in a mini-pilot in a small pond at the premises of Deltares in Delft.


    All the necessary licenses and permits were granted relatively fast and easy, partially because the structure is temporary. Quite a lot of time was spent to the project’s preparation, but the result was that all stakeholders involved were familiar with the project, which benefited the permitting process and public acceptance.



    Detailed design


    The pilot location with floating brushwood mattresses had to be close to shore. The location should be easily accessible by boat and there should be sufficient wave impact to be able to measure the wave-attenuating properties of the mattresses. There should be no cabling or other elements present that would hamper the construction of the mattresses. A location near the Houtribdijk, close to the Houtribsluizen near the city of Lelystad was chosen. The mattresses were not directly reachable from the shore, in order to prevent unauthorized access.

    Mattresses consisted of 4 different layers of brushwood resulting in a total thickness of 40 cm. Reed rhizomes were embedded in each separate layer, with a density of 20 per square meter. The minimum size of the rhizomes was 25-30 cm. In total 48.800 rhizomes were used. The size of the floating mattress complex was 100 by 30 m, as shown schematically in construction figure. Mattresses were constructed on land in three separate units, connected once they were afloat. In the middle of each of the three mattress units open spaces were created with different dimensions. The brushwood part had a total area of 2440 m 2 and a weight of 80.000 kilo. Part 1 and 2 were constructed of 100% Salix alba, part 3 was constructed of 25% Salix alba and 75% Salix viminalis due to a shortage of Salix alba. All the wood was harvested at least 3 years prior to construction.

    Sustainability of designs and their ability to attenuate waves was tested in experiments in the Delta flume (van Steeg & van Wesenbeeck 2011). Six mattresses, each of 4 meter length and 2 meter wide, were constructed in the Delta flume. Two mattresses were placed next to each other and in the length three mattresses were placed behind each other. Total length of the mattresses in the flume was 12 meter. First experiments showed that the main challenge is to create solid anchor points, where willows are not ripped apart due to forces that are executed by waves.

    Operation and Maintenance

    Management and maintenance

    The floating marsh was completed on the 26th of June 2009. By the 1st of October parts of the mattress were starting to sink. Floating capacity of the mattress was enhanced artificially with empty plastic jerry cans. A storm on the 13th of October caused the braided willow branches to disentangle and large parts of the mattresses sank further due to wave impact. Consequently, in December the marsh was removed and the brushwood was transported to a biomass burning power plant. These experiences have generated valuable lessons regarding the management and maintenance of such installations that can be used for further implementation of artificial marshes (floating or sinking).


    The following aspects of the floating marsh were planned to be monitored.

    • Floating ability
    • Development of habitats
    • Biomass production
    • Wave reduction
    • Filter capacity

    Due to rapid sinking of the mattresses, most of these factors could only be monitored qualitatively. The monitoring plan developed for this installation can effectively be adapted for further implementation of reed mattresses, either floating or sinking, as many of the monitoring aspects remain the same.


    • The long-term floating capacity appeared to be too low and the marsh sank within 5 months. This was caused by water absorption of the brushwood and insufficient growth of self-floating reed rhizomes in this period.
    • The first reed growth was observed on July 17th, 3 weeks after completion. In the following month the reed was growing rapidly. Growth was more abundant on the second and third part of the marsh, with a maximum reed height of a little over 2 m and a maximum density of around 15% by September 1st. Grazing of the reed by geese was observed (van Geest et al. 2010).
    • No measurements on wave reduction in the field were made, but during periods with waves a significant reduction of wave energy was visually observed on the sheltered side of the marsh and in the small open areas inside the mattresses. Measurements in the Delta Flume showed that significant wave attenuation can be achieved with floating marshes (van Steeg & van Wesenbeeck 2011).
    • In the sheltered zone in the middle of the mattresses a higher concentration of water birds was observed.
    • During the application of the plastic jerry cans 'significant' amounts of settled mud were observed underneath the mattress. This effect was expected, but could not be quantified.

    Lessons Learned

    • Brushwood in water absorbs a certain amount of water in the course of time. Consequently, brushwood mattresses as tested here keep their buoyancy for at most two to four months. If longer buoyancy is required the design should be adapted, potentially with additional floating devices or by using material with more buoyancy.
    • Flume experiments showed that weakest points of the mattresses were edges and anchoring points. Edges and connections of the brushwood mattresses should be improved in order to better resist wave action.
    • The shape of the floating mattresses and the size and location of the openings should be designed according to ecological requirements, preventing negative shading impacts but optimizing the wave attenuation and sediment capture impacts.
    • Potential impact of geese on the establishment of reed should be assessed before creating reed areas that are partly submerged to minimize failure caused by overgrazing of geese.
    • Establishment of reed by embedding rhizomes in brushwood mats is initially successful but can be improved by also adding sufficient soil so that there is no nutrient limitation for reed growth.
    • Wave attenuation by brushwood mats is effective. For the wave-attenuating effect to be optimal, length, thickness and porosity of the mattresses should be designed in accordance with the local wave conditions.
    • The sinking property of these mats allows application as bottom mats on shallow foreshores. The mats are a suitable method to create (reed) marshes in shallow water.



    Back to Top