COMPARISON OF A COMPUTER-SIMULATED STRATUS-TOPPED BOUNDARY-LAYER WITHAIRCRAFT OBSERVATIONS

To assess the realism of large-eddy simulation (LES) of the stratus-to pped boundary layer and its predicted turbulent structure, we performe d detailed data analyses on a LES (which has a 12.5 m grid size in all three directions), in a manner similar to those used by Nicholls (198 9) on aircraft measurements. The first analysis retrieves the primary convective elements, i.e., the negatively buoyant downdrafts, which ar e driven mainly by cloud-top radiative cooling, through a conditional sampling technique. Comparison shows that the LES of this resolution r eflects most of the observed downdraft features; most of the discrepan cies that exist between the obervations and the LES can be explained b y decoupling of the cloud layer from the underlying flow that exists i n the former but not in the latter. The second analysis shows the vert ical velocity spectrum and its agreement with the measurements. In the third analysis, showing the turbulent kinetic energy budgets, the dis crepancy in the turbulent transport term (i.e., the divergence of the third-moment quantity wEBAR, the turbulent-kinetic-energy flux) betwee n the LES and measurements exists even with such a fine resolution LES . This discrepancy is related mainly to the different behavior in w3BA R between the LES and observations, which may again be associated with decoupling. An advantage of LES over aircraft observations is that th e former can provide three-dimensional flow structure at any instant. In this paper, we examined the instantaneous flow structure and observ ed closed cellular patterns near the cloud top in which updrafts occup y the broad centers and relatively strong downdrafts occur in the narr ow edges. In the intersections of these cell boundaries, there exist w eak downdrafts, consisting of relatively cold and dry air, that are th e most likely origins of the strong downdrafts extending throughout th e mixed layer.