Resolvability of small-scale non-turbulent motions in the boundary layer by numerical models

Mesoscale numerical models apparently do not capture small-scale non-turbulent motions in the boundary layer, and this seems to be the case regardless of the model resolution. The suspected causes are that the models miss necessary physics or miss the initiation of motions by small-scale terrain features and that the small-scale motions are filtered by implicit and/or explicit numerical diffusion. The absence of these submeso motions may underestimate shear-generation of turbulence in models. Also, the dispersion models may not work well with these motions omitted.

Using the hierarchy of nested domains in the numerical mesoscale model WRF we evaluate the effects of numerical diffusion and boundary conditions on the amount of mesoscale variability at resolved scales. As certain previous studies indicate, there seems to be a significant absence of variability at smaller mesoscales using the usual stable model setup. Removal of the implicit and explicit numerical diffusion yields significant growth of the variability at all scales. Although this variability is apparently non-physical, the model spectra agree well with the measurements. We discuss different effects related to the associated upscale or downscale energy cascade at mesoscales.