Understanding the effects of surface roughness transitions on the spatial distribution of surface shear stress and velocity is key to improving predictions of turbulent transport in the atmospheric boundary layer ABL). This is particularly important as a boundary condition for the design of wind energy projects. In the first part of this research, the effects of surface roughness heterogeneity on the ABL were studied in the boundary layer wind tunnel of the Saint Anthony Falls Laboratory SAFL) at the University of Minnesota. We specifically developed a numerical model that accurately predicts, for the first time, the change of wind velocity and surface shear stress downwind of a surface roughness transition. In this context, different surface boundary conditions for large-eddy simulation were tested downwind of a rough-to-smooth surface transition. Results show substantial differences between measured and modeled shear stress using standard boundary conditions based on the direct application of the similarity theory with local fluctuating filtered velocities. The best performance is obtained using the proposed model for estimating the adjustment of the mean velocity and surface shear stress downwind of the transition, while the surface shear stress fluctuations were modeled proportionally to the velocity fluctuations. This improves the prediction of the variance and spectrum of the uctuating shear stress with respect to standard boundary conditions. In the second part of this research, the complex interaction between the ABL and wind turbine wakes) was studied in the SAFL wind tunnel using model wind turbines. The structure and behaviour of the turbulent flow around the wind turbines were characterized under both thermally neutral and stable stratifications. Non-axisymmetric behavior of turbulence statistics in the wake was observed in response to the nonuniformity of the incoming boundary layer flow. Nevertheless, the velocity deficit with respect to the average incoming flow was nearly axisymmetric everywhere except near the surface in the far wake. In the wind farm scenario, results suggest that the turbulent flow can be characterized into two main regions. The first, located below the turbine top tip height, has a direct effect on the performance of turbines. Here the mean flow statistics appear to reach an equilibrium as close as 3-4 turbines downwind of the first turbine. In the second region, which is located immediately above the turbine top tips, flow adjustment is slower. Here, two distinct layers were found: an internal boundary layer where the flow starts to adjust to the new farm surface conditions, yet is still affected by the upwind flow characteristicsï¼› and an equilibrium layer, where the flow statistics are fully adjusted to the wind farm conditions. Our results also show that wind turbine wakes reduce the mean surface heat flux, where a large wind farm implied the most significant change. This observation points to the necessity of new parameterizations for large scale models.