9.4
Building and terrain effects in a mesoscale model
Ted Yamada, Yamada Science & Art Corporation, Santa Fe, NM
Vu Thanh Ca et al. (1999) considered building effects in their mesoscale model where horizontal grid spacing of 10 m was used over a computational domain of 500 m x 500 m. Brown et al. (1998) embedded a two dimensional CFD (Travis et al., 1994) in a three-dimensional mesoscale model (Yamada and Bunker, 1988) to simulate the flow and concentration distributions around two 2-D buildings.
None of mesoscale models explicitly consider building and terrain effect simultaneously. Terrain effect extends typically more than 10 km so that a computational domain should be at least in the order of 10s km. Consequently, a typical horizontal grid spacing used in a mesoscale model is in the order of 1 km or greater. On the other hand, a horizontal grid spacing of 1-10 m would be needed to resolve the building effect in an urban area.
It is not practical to employ a horizontal grid spacing of 1 m for an entire mesoscale study area of over 10 km. We will explore a feasibility of nesting several grids in a mesoscale domain where horizontal grid spacing reduces from 1 km to 1 m. A nesting grid method significantly reduces computational load, yet maintains accuracy at a high resolution area. We employ a two-way nesting method where outer grids provide boundary conditions to the inner grids and the values previously computed in the outer grids are replaced by the newly computed values in the inner grids. In this way, terrain influence is transmitted to the inner grids and building effect is transmitted to the outer grids.
Session 9, Advanced modeling techniques for dispersion on all scales
Wednesday, 12 January 2000, 8:30 AM-10:00 AM
Previous paper Next paper