The air flow in the lower atmosphere on the spatial scale of the modern wind turbines is studied. Because wind turbines are nowadays often taller than 100 m, the validity of current analytical and numerical atmospheric models has to be evaluated and more knowledge about the structure of the atmospheric boundary layer at those heights has to be acquired. A new long-range wind lidar was placed next to well-instrumented meteorological masts in the west of Denmark and in northern Germany and measured the wind speed and direction up to 2000 m and was compared with wind-speed measurements from the meteorological masts. The Høvsøre site is characterized by a transition from the flow over sea, which has a low surface roughness, to flow over land, which has a much higher surface roughness. The internal boundary layer that forms after this transition, was characterized by both the upstream and the downstream stability. Modelling the mean climatological wind speed with a 3-layer interpolation scheme gave good results, both in neutral conditions and when including other stability conditions. The constants in the model were slightly adjusted based on comparison with other studies and a numerical model. A mesoscale numerical model was used to simulate the flow at Høvsøre for four weeks during autumn 2010. The wind profile did not have enough vertical shear in the lower part of the PBL and had a negative bias higher up in the boundary layer. In the grid points after the shoreline the wind speed near the surface and the friction velocity had a bias, which were related to the change in surface roughness. A higher-order boundary-layer scheme represented the wind profile of the westerly flow over sea better, while a first-order scheme modelled the flow from the east with low-level jets better. The wind profile shape and the negative wind speed bias at larger heights were not improved when a different synoptic forcing and a different vertical resolution were used in the model. The effect of baroclinity was explored for the two sites. The surface geostrophic wind, the gradient wind and the thermal wind were derived from simulations with a mesoscale model. In both locations the thermal wind up to 970 m was approximately Gaussianly distributed with a standard deviation of three m s−1 and the thermal wind vector varied seasonally due to temperature differences between sea and land. The wind veer was particularly sensitive to baroclinity. The variation of the resistance law constants in neutral, baroclinic conditions was approximately the same as in experiments that where assumed to be barotropic; part of the variation was explained by baroclinity showing the importance of including this effect when studying boundary-layer winds.
DTU Wind Energy PhD 34(EN); DTU Wind Energy PhD-0034