High-resolution mesoscale simulations on the role of shallow and deep convection on dust emission and transport in a desert area

Recent studies on dust emission and transport from arid regions have been concerning the micro-scale atmospheric processes in order to incorporate them as a subgrid-scale effect in large-scale numerical prediction models. In order to develop a better parameterization of these subgrid-scale processes, we need to comprehend the physical aspects of those micro-scale processes which include thermals, boundary-layer turbulence and convection, and deep cumulus convection. For this purpose, we have performed a series of high-resolution simulations explicitly resolving organized convective motions in a mesoscale domain to investigate the dynamical processes of dust emission and transport induced by shallow and deep cumulus convection under fair-weather conditions in a midlatitude desert area. The simulation set employs a horizontal grid size of 500m and examines the sensitivity of the dust dynamics to vertical wind shear, upper-level wind speed, and moist convection. The simulated results qualitatively capture the diurnal variation of the dust layer observed by lidar in a desert area in northern China. Shallow convection plays a primary role in vertically mixing dust within the boundary layer. If deep cumulus convection also comes into play, mass concentration of dust increases not only within the boundary layer but also in the free troposphere. A coupled effect of dry and moist convection is important because convection is more enhanced with the coupled effect than without moist processes and in addition transports upper-level higher momentum down to the surface, intensifying surface winds and hence dust emission. A wind speed exceeding the threshold for surface dust emission is necessary at the upper levels of the daytime boundary layer for the surface wind enhancement. Although the amount of dust lifted is much smaller during a single diurnal cycle of fair weather than associated with a single case of synoptic disturbances, the total amount of dust emission due to fair-weather processes may not be neglected in a longer time-scale.

We have also performed higher-resolution simulations by further decreasing the grid size to the order of 100 m: large-eddy motions of shallow and deep convection are explicitly resolved in the mesoscale computational domain. The spatial variability of the vertical velocity reproduces a spectral drop-off seen in a turbulence inertial subrange, which indicates that the simulation can be regarded as a large-eddy simulation. Although the overall features of convection development and dust transport appear to be similar to the results obtained with the coarser-resolution simulations and the times of the initial dust emission between the coarse- and fine-resolution simulations correspond well with each other, the onset and the gradual development of shallow convection and the transition to deep cumulus convection are sufficiently resolved. This is a significant improvement from the coarser-resolution simulation which on the other hand produces an abrupt development of deep convection. A significant difference between the coarse- and fine-resolution simulations is found in the column amount of dust floating in the air. The results of the higher-resolution simulations are further compared with those of low-resolution simulations with a grid size of O(1 km) in terms of velocity spectra. The subgrid-scale mixing due to a turbulence parameterization seems to be effective in the cases with the grid size of 500 m and lower. These results suggest that a parameterization of subgrid-scale turbulence mixing that transports dust aerosols in the smallest scale needs to be revised.