Large-eddy simulations of thermally forced circulations in the convective boundary layer. Part I: A small-scale circulation with zero wind

We have conducted large-eddy simulations (LES) of the atmospheric boundary layer with surface heat Rux variations on a spatial scale comparable to the boundary layer depth. We first ran a simulation with a horizontally homogeneous heat flux. In general the results are similar to those of previous large-eddy simulations. The model simulates a field of convective eddies having approximately the correct velocity and spatial scales, and with the crucial property that kinetic energy is transported vigorously upwards through the middle levels. However. the resolved temperature variance is only about half what is observed in the laboratory or the atmosphere. This deficiency which is shared by many other large-eddy simulations has dynamic implications. particularly in the pressure/temperature interaction terms of the heat flux budget. Recent simulations by other workers at much higher resolution than ours appear to be more realistic in this respect. The surface heat flux perturbations were one-dimensional and sinusoidal with a wavelength equal to 1.3 times the boundary-layer depth. The mean wind was zero. Results were averaged over several simulations and over time. There is a mean circulation. with ascent over the heat flux maxima (vertical velocity -0.1~~) and descent over the heat flux minima. Turbulence is consistently stronger over the heat flux maxima. The horizontal velocity variance components (calculated with respect to the horizontal average) become unequal, implying that convective eddies are elongated parallel to the surface heat flux perturbations. A consideration of the budgets for temperature and velocity suggests several simplifying concepts,