Other reports in this collection
Turbulent motions affect all exchanges of heat, water, momentum and chemical constituents between the surface and the atmosphere, and these motions are also responsible for the mixing processes inside the atmospheric boundary layer. Consequently, the turbulent motions impact on the formation and existence of fog and boundary-layer clouds such as cumulus, stratocumulus and stratus. Cumulus is typically found in fair weather conditions over land and sea, while layers of stratocumulus and stratus can be found in subsidence areas such as the anticyclonic areas in the eastern part of the sub-tropical oceans, or the polar regions.
The atmospheric boundary layer with clouds is typically characterised by a well-mixed sub-cloud layer of order 500 metres, and by a more extended conditionally unstable layer with boundary-layer clouds up to 2 km. The latter layer is very often capped by a temperature inversion. If the clouds are of the stratocumulus or stratus type, then conservative quantities are approximately well mixed in the cloud layer. The lowest part (say 10%) of the sub-cloud layer is known as the surface layer. In this layer the vertical gradients of variables are normally significant, even with strong turbulent mixing. Physical problems associated with the surface layer depend strongly on the type of surface considered (such as vegetation, snow, ice, steep orography) and are treated in the corresponding sub-sections. We note that the surface characteristics do impact on the formation of boundary-layer clouds, because of the turbulent mixing inside the boundary layer.
Atmospheric models have great difficulty in the proper representation of turbulent mixing processes in general. This also impacts on the representation of boundary-layer clouds. At present, the underprediction of boundary-layer clouds is still one of the most distinctive and permanent errors of AGCMs. This has been demonstrated through AMIP intercomparisons by Weare and Mokhov (1995) and also by Weare (2000b). It has a very great importance, because the albedo effect of these clouds is not compensated for by a significant greenhouse effect in both clear-sky and cloudy conditions.
The persisting difficulty in simulation of observed boundary layer cloud properties is a clear testimony of the still inadequate representation of boundary-layer processes. A variety of approaches is followed, ranging from bulk schemes in which the assumption of a well-mixed layer is made a priori, to discretised approaches considering diffusion between a number of vertical layers. Here the corresponding diffusion coefficients are being computed from dimensional analysis and observational data fitted to it. The use of algorithms based on the prognostic computation of turbulent kinetic energy and higher-order closure hypotheses is also becoming more common and new schemes continue to be proposed (Abdella and McFarlane, 1997). A critical review and evaluation of boundary-layer schemes was recently made by Ayotte et al. (1996). They found that all schemes have difficulty with representing the entrainment processes at the top of even the clear boundary layer.
Important and still open problems include the decoupling between the turbulence at the surface and that within clouds, the non-local treatment of semi-convective cells (thermals) that can transport heat and substances upward, the role of moist physics, and microphysical aspects (Ricard and Royer, 1993; Moeng et al., 1995; Cuijpers and Holtslag, 1998; Grenier and Bretherton, 2001). Several studies (Moeng et al., 1996; Bretherton et al., 1999; Duynkerke et al., 1999) show that column versions of the climate models may predict a reasonable cloud cover in response to observed initial and boundary conditions, but have more difficulty in maintaining realistic turbulent fluxes. The results are also very dependent on vertical resolution and numerical aspects (Lenderink and Holtslag, 2000). This points to the need for new approaches for boundary-layer turbulence, both for clear-sky and cloudy conditions which are not so sensitive to vertical resolution (see also contributions in Holtslag and Duynkerke, 1998).
|
Other reports in this collection |
|