The experimental identification of a new boundary layer in hard convective turbulence by Belmonte, Tilgner and Libchaber (1994) using spectral cutoffs has been confirmed numerically using DNS by Kerr, Brandenburg and Herring. But the conclusion that it is related to the usual ways of determining a velocity boundary layer is not confirmed. In the Display Image (4 K) figure of heights versus Rayleigh number from the simulations, the new boundary layer is constant with Rayleigh number and interior to all the other measures of velocity and thermal boundary layers, which decrease rapidly with Rayleigh number. Display Image (53 K) Color contours with arrows for velocity show a predominant recirculating pattern in one direction and convergence zones in the other direction. Why is confirmation of DNS by laboratory experiments important in the atmospheric context?
A major goal of HPCC (the High Performance Communication and Communications program) is to use direct numerical simulations (DNS) for studying the atmospheric boundary layer. The objective of these simulations would not be to reach the Reynolds and Rayleigh numbers of the atmosphere, but allow the natural instabilities that might be suppressed by Large-Eddy Simulation methods. Since we will be attempting to extrapolate scaling laws from these moderate Reynolds number simulations to atmospheric conditions, we need confirmation that such a procedure is applicable. One way to test this premise is to compare our DNS with laboratory experiments where the scaling laws are known to much higher Reynolds numbers than can be achieved by the DNS. Recently, there has been strong experimental interest in the scaling of the velocity boundary layer in Raleigh-Benard convection. This scaling should affect parametrizations of the heat flux not only in the laboratory experiments, but also the atmosphere. We are actively pursuing this angle for understanding thermal convection over the tropical ocean. It is important to verify that the procedure of taking moderate Reynolds number DNS and extrapolating to higher Reynolds numbers works by direct comparisons with the experimental measurements of boundary layer scaling. One notable recent success is reproducing the scaling of a boundary layer determined by spectral cutoffs. The value of DNS becomes apparent when not only are we able to reproduce the laboratory measurement, but are able to interpret it by means beyond current measurement techniques. We also have preliminary results consistent with changes in Display Image (112 K) observed in the laboratory. More information on convection with periodic sidewalls and Pr=0.7 is available.
