This dissertation presents an observational and a numerical study of sea/land-breeze circulations as well as a theoretical study on the boundary-layer turbulence parameterization. Data from the Kennedy Space Center Atmospheric Boundary Layer Experiment have been used to examine the three-dimensional structure of the mesoscale Atlantic sea/land-breeze and small-scale river-breeze circulations frequently occur over the area of Kennedy Space Center/Cape Canaveral (KSC/CC) where there exits irregular topographic features such as rivers and lagoons. Detailed characteristic features over a diurnal cycle of the circulations onset time, strength, depth, propagation speed, and both seaward and landward extensions, are documented. Some boundary-layer characteristics related to atmospheric diffusion including atmospheric stability, depth of the thermal internal boundary layer and its evolution with time, and turbulence mixing are also discussed;A three-dimensional numerical mesoscale model is used to study the effects of synoptic forcing on the sea/land-breeze circulation at KSC/CC area. The sensitivity of mesoscale features to the large-scale background flow is explored by examining the low-level convergence field, location, strength, and timing of maximum vertical motion; diurnal variation of depth of inflow layer and the thermal internal boundary layer; and the rotation of wind vector;Both of these studies may lead to not only a improved forecast of local weather and atmospheric diffusion of rocket exhausts at KSC/CC area, but also to a better understanding of the physical mechanism of the thermally induced circulations, such as their dependence on the three-dimensional coastal effects and the synoptic forcing;A new scheme for parameterizing the turbulent stress in terms of the mean wind profile is also presented in this dissertation. This new parameterization scheme is based on the balance of turbulence kinetic energy and is valid in a deeper layer than the constant-flux layer where the previous schemes all fail because of the violation of the fundamental assumption of constant turbulent fluxes. This new scheme may have direct impact on boundary-layer and mesoscale modeling because it may improve the accuracy of solutions and also reduce the computation time by allowing a minimum number of vertical nodes needed to capture the sharp gradients near the low boundary. (Abstract shortened by UMI.)