roughness patches for ABL flow
 

I am trying to do ABL simulation with roughness patches (length, z0) using wall functions. For a suburban roughness of z0=0.42, the requirement that the first cell be larger than ks (~20z0 = 8.4m) results in cell height of about 17m. For urban patch it would be 34m. Can I trust the result I get using this cell size ? I want to know if this option is feasible at all before pursuing the other option of putting blocks to account for roughness. I have seen one paper by Wang (2007) that does a 2D simulation without using blocks (I think through specifying shear stress but I don't understand it).
thank you