Laboratory experiments on Rayleigh-Benard convection with no imposed mean flow have shown that at sufficiently high Rayleigh numbers a mean horizontal flow u(z) is spontaneously generated. This flow is maintained against viscous dissipation by the countergradient momentum transport of the Reynolds stress uw associated with the tilted plumes. The shear of the mean flow ∂u/∂z at mid-depth may be of either sign, depending upon initial conditions. However, for either sign, the tilt of the plumes is always such that horizontal momentum is transported up the mean gradient in the interior. In similar experiments, but with an externally imposed shear flow, it is found that for a sufficiently large shear and sufficiently large Rayleigh number, the tilt of the plumes is determined by the imposed shear. For either sign of shear the plumes tilt in such a way as to transport momentum up the mean gradient. Sufficient data now exists to express the momentum flux as a function of the Rayleigh number and the imposed shear.
In the atmosphere measurements of vertical transport of horizontal momentum, counter to the mean gradient, include those in the sub-cloud layer under fair weather cumulus, and in deep cumulonimbus convection in several squall lines observed during GATE.
Although the parameterization of heat and moisture flux by cumulus convection has been successfully incorporated into numerical weather prediction models, momentum flux has generally not been so included. We attempt to include such a parameterization of the vertical momentum flux by convection, using laboratory measurements as a guide. In this application the conventional Rayleigh number is replaced by an atmospheric parameter which is proportional to the vertical gradient of the equivalent potential temperature. The momentum flux by the convective clouds is then parameterized as a function of the modified Rayleigh and Reynolds numbers.
The results from a control experiment without the inclusion of this parameterization are compared with an experiment where it is included. Of specific interest to the study are the errors in the trade winds of the Atlantic and Pacific Oceans and the low level monsoon circulation over the Indian Ocean. This formulation is shown to reduce the systematic errors of the low level winds over the central Pacific Ocean quite dramatically by over 10 ms-1. However the reduction of errors over tropical land areas, especially around regions of elevation exceeding 500mb, are smaller.
View full abstract