Research Direction

We are trying to develop a two-scale  boundary layer model of unresolved circulation in the planetary boundary layer(PBL) in a manner which is independent of the size of the underlying grid. Such an approach is advantageous for advanced weather prediction enterprise which rely on a range of models (each one suitable for a particular scale), as it ensures consistency among parametrization irrespective of the underlying model. The two scale boundary layer model would provide a unified framework for PBL modeling for global ,regional as well as cloud scale modeling since it is independent of the size of the underlying grid of the numerical model.

Current approaches to PBL modeling is based on methods where the central idea is that the turbulent length scale is much smaller than the scale of motions resolved by the underlying grid. This is the reason why the turbulence models are usually supplemented with local mixing rules which help insure that the resolved scale flow remains stably stratified and smooth in the vertical. These schemes  are usually explicitly or implicitly tied to the vertical grid structure. However there are no models which consistently does this tie-up as a function of horizontal and vertical resolution.

We are developing a two scale turbulence kinetic energy closure scheme. This will contain a large scale TKE which contains the energy of unresolved eddies that scale with the depth of the PBL .The other TKE is that of the small scale unresolved eddies which scale with the smallest grid increment in the vertical direction. Mathematically it can be shown that a small eddy diffusivity can be defined depending on the square root of the small scale TKE and a large eddy velocity scale can be defined depending on the square root of the large scale TKE. The evolution equations of the small and large scale TKE are written such that the dissipation of large scale TKE is the source of small scale TKE thus formalizing the turbulence cascade.

However challenges remain. Firstly we have to establish proof of these concepts for example for the idealized convective layer. Secondly blending functions need to be developed which allow the scheme to work as the PBL turbulence is increasingly resolved. Currently I am trying to work out a simple single column model for temperature,water vapor and other important variables. The idea is to try to implement the two TKE scheme in the single column model to test it before implementing it in ICON.