Simulating the diurnal cycle of the atmospheric boundary layer using large eddy simulation with vertical adaptive mesh refinement

The diurnal evolution of the atmospheric boundary layer (ABL) plays an important role in weather and climate systems. Many numerical studies break the diurnal cycle up into its two primary components, the daytime convective period and the nocturnal stable period. This is especially prevalent in large-eddy simulation (LES) studies were simultaneously representing the diurnal cycle's two distinct regimes creates domain size and resolution problems. During the daytime, when positive buoyancy forces result in a rapidly growing boundary layer with highly energetic, large-scale turbulence the domain size must be large enough to capture these motions. In contrast, at night negative buoyancy forces suppress turbulence resulting in a much lower boundary layer height and significant small-scale turbulent fluxes necessitating a finer grid resolution. While both regimes have been modeled successfully in independent simulations, most diurnal cycle LES studies under resolve the nocturnal boundary layer in order to capture the large-scale features of the convective boundary layer. Here, LES is combined with adaptive mesh refinement in the vertical direction to achieve a more efficient and robust simulation of the diurnal ABL. Vertical grid spacing is adjusted with relocation-type (r-type) refinement to provide increased vertical resolution in the boundary layer without adding the computational overhead of additional points. Simulations of the 24-hour Wangara case (days 33/34) demonstrate that LES with AMR can successfully adapt in space and time to changing turbulent length scales providing increased resolution within the boundary layer and improved results compared to simulations with a static grid and the same total number of grid points. During the daytime convective period, increased resolution in near surface regions leads to better-represented first and second order statistics. During the nocturnal period, AMR simulations cluster grid points within the boundary layer improving the representation of stratified turbulence. While the representation of the nocturnal boundary layer is improved, results highlight the importance of employing sufficient horizontal resolution during stratified periods.