Improvement on the parameterization for atmospheric boundary layers in the JMA global NWP model

Atmospheric boundary layer (ABL) is an important object of a numerical weather prediction (NWP). The Japan Meteorological Agency (JMA) global NWP model employs a first order turbulence closure scheme to handle effects of atmospheric turbulence. The scheme determines a vertical diffusivity both within and above ABL based on the local Richardson number.

In many NWP models, so-called non-local approach is often applied to express unstable situations. Since the JMA model does not treat a non-local mixing explicitly, a diffusion coefficient within ABL is limited by a minimum value to avoid insufficient mixings. Introduction of a non-local diffusion scheme to the model makes turbulent mixings more efficient under unstable condition and modifies diurnal variations in surface layer.

In the column model intercomparison experiment of GEWEX Atmospheric boundary layer study (GABLS), on the contrary, a single column version of the model represents too strong turbulent mixing in the stable ABL case. It is also suggested that the excessive mixing resulted from a large minimum of diffusivity and overestimation of mixing lengths in the stable condition. The modification to these defects found in the experiment reduces a nocturnal strong wind bias of the model over land surface.