Implementation and Testing of Advanced Surface Boundary Conditions Over Complex Terrain in A Semi-idealized Model

In numerical prediction models, the interaction between the Earth's surface and the atmosphere is typically accounted for in terms of surface layer parameterizations, whose main job is to specify turbulent fluxes of heat, moisture and momentum across the lower boundary of the model domain. In the case of a domain with complex geometry, implementing the flux conditions (particularly the tensor stress condition) at the boundary can be somewhat subtle, and there has been a notable history of confusion in the CFD community over how to formulate and impose such conditions generally. In the atmospheric case, modelers have largely been able to avoid these complications, at least until recently, by assuming that the terrain resolved at typical model resolutions is fairly gentle, in the sense of having relatively shallow slopes. This in turn allows the flux conditions to be imposed as if the lower boundary were essentially flat.

Unfortunately, while this flat-boundary assumption is acceptable for coarse resolutions, as grids become more refined and the geometry of the resolved terrain becomes more complex, the appproach is less justified. With this in mind, the goal of our present study is to explore the implementation and usage of the full, unapproximated version of the turbulent flux/stress conditions in atmospheric models, thus taking full account of the complex geometry of the resolved terrain. We propose to implement the conditions using a semi-idealized model developed by Epifanio (2007), in which the discretized boundary conditions are reduced to a large, sparse-matrix problem. The emphasis will be on fluxes of momentum, as the tensor nature of this flux makes the associated stress condition more difficult to impose, although the flux conditions for heat and moisture will be considered as well.

With the resulotion of 90 meters, some of the results show that the typical differences between flat-boundary cases and full/stress cases are on the order of 10%, with extreme cases reaching as high as 30% based on typical disturbance wind speeds. And this difference dropping by a factor of six between grid spacings of 90 meters and 240 meters. It would thus appear that the need to apply the full stress condition is limited to relatively high-resolution modeling, with grid spacings on the order of 250 meters or less.