In 2007/08, a study was undertaken on the sediment dynamics in shallow Lake Markermeer (the Netherlands). Firstly, sediment characteristics were determined at 49 sites in the lake. Parameters such as median grain size and loss on ignition showed a spatial as well as water depth related pattern, indicating wind-induced sediment transport. Highly significant correlations were found between all sediment parameters. Lake Markermeer sediment dynamics were investigated in a sediment trap field survey at two permanent stations in the lake. Sediment yields, virtually all coming from sediment resuspension, were significantly correlated with average wind speeds, though periods of extreme winds also played a role. Sediment resuspension rates for Lake Markermeer were high, viz. on average ca. 1,000 g m -2 day -1 . The highly dynamic nature of Lake Markermeer sediments must be due to the overall shallowness of the lake, together with its large surface area (dynamic ratio = [H(area)]/[average depth] = 7.5); wind-induced waves and currents will impact most of the lake's sediment bed. Indeed, near-bed currents can easily reach values [10 cm/s. Measurements of the thickness of the settled ''mud'' layer, as well as 137 Cs dating, showed that long-term deposition only takes place in the deeper SE area of the lake. Finally, lake sediment dynamics were investigated in preliminary laboratory experiments in a small ''micro-flume'', applying increasing water currents onto five Lake Markermeer sediments. Sediment resuspension started off at 0.5-0.7 cm/s and showed a strongly exponential behaviour with respect to these currents.

Lake Markermeer is a large (680 km 2 ), shallow body of water in the middle of the Netherlands, with a mean water depth of 3.6 m. One of the major problems in the lake is its decreasing ecological value which is, among other reasons, caused by a gradual increase of suspended sediment concentration and associated increase of light attenuation in the water column. A thorough understanding of fine sediment dynamics in the lake is a prerequisite for solving this problem. This paper addresses the 3D nature of near-bed sediment dynamics in Lake Markermeer, based on data sampled from a 1-month field experiment in autumn 2007. The campaign involved the collection of 71 bed samples across the lake. At each location, dual-frequency echo soundings were carried out to assess the thickness of the silt layer, and sediment concentration throughout the water column was measured with an Optical Backscatter Sensor (OBS). Moreover, 2-week time series of wave height, water level, current velocities, and near-bed sediment concentration were collected at a single location. The time series of sediment concentration were measured with a regular OBS and an Argus Surface Meter IV (ASM). During the measurement period, flow velocities ranged between 2 and 15 cm/s, wave heights up to 1.2 m were observed and turbidity levels varied between 40 mg/l to more than 300 mg/l. The ASM data generally showed uniform concentration profiles. However, profiles with steep concentration gradients near the bed were found for wave heights above 0.5 m. The field experiments further revealed pronounced 3D structures near the bed during discrete storms. The results are generalized for a wider range of conditions and across the full water depth through application of a 1DV point model, using a two-fraction representation of the grain size distribution. The fine and coarse fractions are found to resuspend rapidly for wind speeds above 5 m/s and 10-12 m/s, respectively, forming a uniform concentration profile if these wind conditions persists. High-concentration (∼1 g/l) layers near the bed, containing the coarse sediment fraction, only occur at the onset and towards the end of a storm, when wind speed changes rapidly. It is under these conditions that horizontal gradients in layer density or thickness can transport considerable fine sediment. This transport provides an additional mechanism for the infill of, for instance, silt traps and navigation channels.

The performance of three different hydrodynamic modelling packages is compared in this study, namely Delft3D, Delft3D FM (both developed by Deltares) and MIKE 21 FM (developed by DHI). Delft3D and MIKE 21 FM are internationally known software packages while Delft3D FM (formerly known as D-Flow FM) is a relatively new package. The models use structured approaches (Delft3D), unstructured approaches utilising triangular and quadrilateral elements (MIKE 21 FM) and unstructured approaches utilising elements ranging from linear to six sided (Delft3D FM). Models of Western Port, Australia, were developed using the three different packages to allow a comparison of performance and to determine if there are any differences in using structured versus unstructured approaches. Model performance has been assessed based on model calibration, representation of channel flows and computational efficiency. Despite the inherent differences in the grid configuration and the implementation of the numerical schemes between structured and unstructured approaches, both approaches have been shown to be able to accurately predict hydrodynamic conditions in a complex estuarine environment. The unstructured approach was found to be the most computationally efficient both when run on multiple cores (MIKE 21 FM was the most efficient) and when run on a single core (Delft3D FM was the most efficient).