Author(s): Brett F. Sanders; Robert D. E.
Linked Author(s): Brett F. Sanders
Keywords: Flood inundation modelling; Flow resistance; Digital terrain model
Abstract: Hydrodynamic simulation of flooding over natural topography essentially requires two spatially distributed parameters: ground elevation data that defines the geometry of the study site and a drag coefficient that budgets resistance caused by bed shear and form drag. It is conventional to scale bed shear by a Manning/Strickler, Chezy, or Darcy-Weisbach resistance coefficient that is based on the roughness of the ground surface. Form drag, on the other hand, is associated with obstructions such as vegetation and buildings. High-resolution Digital Terrain Models (DTMs) now make it possible to grid terrain with ~10 cm vertical accuracy and ~1 m spatial resolution, revealing earth surface texture or bed forms not previously seen in less accurate and coarser resolution DTMs. In this paper, we show that flow resistance budgeted by a hydrodynamic flood inundation model is sensitive to bed-forms resolved by the computational grid, or “mega-roughness.” A model prediction of flooding over smooth terrain budgets less resistance than a model prediction of flooding over uneven terrain when the same resistance coefficient is used for both. The effect is necessarily grid dependent because more surface texture is resolved as the grid is refined. The implication is that resistance parameters (e. g. Manning n) could be scale dependent, with relatively smaller parameter values being adopted with grid refinement. There is an obvious analogy between this and the eddy viscosity concept adopted for large eddy simulation.
Year: 2007