The earth's surface is permanently exposed to the atmosphere and accordingly to strong wind forces in many regions. Aerodynamic entrainment, transport and redeposition of sand, soil or snow are able to considerably reshape the surface morphology and influence the environment in areas ranging from deserts to polar regions. Even in moderate climate zones, entrainment and transport of dust, particulate matter and pollen or seeds by wind may have a strong impact on the local atmosphere and vegetation. Many of these processes exert negative influences on our sensitive natural environment. Land degradation, desertification or dust storms, increased particulate matter concentrations in the atmosphere or reduced accumulation of snow in arid regions are just a few examples of the impacts of wind erosion. Vegetation on the ground can provide an efficient sheltering effect against wind erosion. Plants influence sediment erosion mainly by the following four mechanisms: by reducing the surface exposed to the wind, by trapping particles in motion, by local stress concentration and by absorbing momentum from the flow. The latter results in lower surface shear stress on the ground beneath the plant canopies. The peak of the surface shear stress is responsible for the onset of erosion and the spatial mean is commonly used to estimate particle mass fluxes. To quantify the sheltering effect of vegetation, a method called shear stress partitioning has been extensively investigated in the past. This method determines the fraction of the total fluid stress on the entire canopy acting directly on the substrate surface. However, previous studies have limitations: they were either field-based, mainly using live plants, with the limitation that wind conditions could not be controlled, or from wind tunnels using rigid and non porous plant imitations, that poorly reflect the aerodynamic behaviour of live vegetation. This study takes a new approach, performing shear stress partitioning experiments in a controlled wind tunnel environment to systematically quantify the sheltering effect of live, flexible and porous plants. Subsequently, the data was used to test and improve a theoretical model that predicts the stress partition for vegetation canopies. This dissertation is divided into four sections. The main results of each section are discussed in this thesis and have also been published as one conference (Chapter 2) and three journal articles (Chapter 3-5). In Chapter 2, the flow conditions produced in the wind tunnel over live vegetation canopies were investigated to identify the suitability of the boundary-layer flow for these new investigations of shear stress partitioning. Flow characteristics like vertical Reynolds stress and integral length scale profiles and power spectral densities were determined from two-component hot-film anemometry measurements. The results were in good agreement with established literature, suggesting that well developed boundary-layers over live vegetation canopies can be generated in this wind tunnel. In Chapter 3, the experimental setup and the building, calibration and testing of the measurement technique for measuring surface shear stress in the wind tunnel are presented. The experimental setup consisted of wooden boards in which the live plants, grown in plastic tubes, were arranged in staggered rows. The surface shear stress sensors (Irwin sensors) and the required multi-channel pressure scanner were custom designed and built. Accuracy tests verified that reliable surface shear stress measurements with an average accuracy of about ±5% can be performed when using a universal calibration function for all Irwin sensors built for this study. The surface shear stress distribution around a single wall-mounted rectangular block was measured as a test case and can serve as high-resolution validation data for CFD simulations. The surface shear stress distributions on the ground beneath the different densities of live plant canopies were measured with previously unmatched high spatial and temporal resolution, the results of which are presented in Chapter 4. Vertical velocity profiles were measured with a two-component hot-film anemometer to determine the total stress above the canopy as well as additional flow characteristics. For comparison, similar experiments were performed with rigid blocks as substitutes for the plants to systematically investigate the influence of the plants' flexibility and porosity on their sheltering effect against sediment erosion. Several distinctive differences in the sheltering effect of live plants and rigid blocks were found: (i) Flow speed-up around the blocks caused higher peak surface shear stress than in experiments with plants. (ii) The sheltered areas in the lee of the plants are significantly narrower and longer with higher surface shear stress than those found in the lee of the blocks. (iii) The streamlining behaviour of the flexible plants results in a decreasing sheltering effect at increasing wind speeds. (iv) Turbulence intensity distributions close to the ground suggest a suppression of horseshoe vortices in the plant case. Another important result is that the percentage of time when a particle entrainment threshold surface shear stress value is locally exceeded is found to be a useful parameter for determining local erosion and deposition rates. In Chapter 5, a shear stress partitioning model (Raupach 1992) was tested against the measured data. The model allows the prediction of the total shear stress on the entire canopy as well as the peak and average shear stress ratios. This study is the first, to systematically investigate the models ability to account for shape differences of various roughness elements. The model can predict the general difference between the plant and the block experiments correctly, although the model limitations were clearly revealed and are discussed in this chapter. The model constant c, relating the size of an effective shelter area and volume to flow parameters and which was poorly specified prior to this study, was found to have a value of about c = 0.27. Values for the model parameter m, which relates the peak surface shear stress to the spatial average shear stress, are difficult to determine because m was found to be a function of the roughness density, the wind velocity and the roughness element shape. A new, more physically based parameter a referred to as the peak-mean stress ratio is suggested as a substitute for m which is solely a function of the roughness element shape. According to this, values for a are much easier to determine than values for m. As a result, a method to identify values for the new a-parameter for different kinds of roughness elements is presented.