Examining the nesting-LES capability using the WRF model

Large-Eddy Simulation (LES) explicitly resolves energy-containing turbulent motions that are responsible for most of the turbulent transport, and hence is treated as a turbulence-resolving model. It has been used intensively in the PBL and engineering fluid dynamics communities to examine the detailed structure of turbulence and its statistics behavior. However, most of the LES applications, particularly for the PBL, have been limited to idealized turbulent flows, i.e., PBLs over horizontally uniform surfaces or periodic striplike heterogenous surfaces where periodic boundary conditions can be applied in the horizontal directions.

The use of periodic boundary conditions prevents us from simulating realistic meteorological flows over complex terrain or land use. One way to tackle complex turbulent flows in weather forecast models is to explicitly resolve both turbulent and mesoscale motions of interest through nesting. With the increasing computer power, this multi-scale nesting approach for turbulence is becoming feasible. But first, we need to examine the feasibility of the nesting LES technique. Unlike weather events, turbulence extends throughout the horizontal domain within the boundary layer, even at the nesting boundaries. Any numerical noises due to specified boundary conditions would quickly spread into the whole domain via turbulent mixing. With the specified velocities at the boundaries, the mass continuity may not satisfy near the nesting boundaries. Thus, nesting LES simulations place a much greater demand on the capability of the information exchanges across the nesting boundaries.

In this study, we examine the horizontal nesting capability with two LES domains under identical forcing and environmental conditions using the WRF model. The inner-domain LES covers just a portion of the horizontal extent of the outer-domain LES, with a horizontal grid size three times smaller than that in the outer-domain LES. The outer-domain LES uses a periodic boundary condition in x and y (like conventional PBL LESs), while the inner-domain LES uses a specified lateral boundary condition based on the outer-domain flow. In this talk, we will show that with some proper modifications to the WRF model, both LES domains can produce the same statistics, as they should under the same large-scale forcing and environment. The two turbulent flows can blend in smoothly across the nesting boundaries and interact properly. The results show promising for applying the nesting WRF-LES to study real-world PBL problems, e.g., PBLs over complex terrain or multi-scale interaction between PBL turbulence and mesoscale motions.