Evaluating 1 and 2D Dimensional Models for Floodplain Inundation Mapping
Abstract
Recent work has illustrated the potential of two dimensional finite element codes for modelling flow in meandering compound channels. Such models have been shown to offer a number of advantages over alternative one (Cunge et al., 1980; Fread, 1985) and two (Zeike and Urban, 1981) dimensional finite difference schemes due to their ability to represent complex topography with a minimum number of computational nodes and the potential accuracy of the finite element method (Huyakorn and Pinder, 1983). Whilst the potential utility of this class of scheme for river flow applications is clear, insufficient model validation remains a major constraint on the development of practical engineering tools. In particular, current studies largely compare model predictions to real observations on the basis of bulk flows (discharges) at the reach downstream boundary (see for example Gee et a!., 1990). This result is typically achieved through calibration of the model friction parameters to replicate this downstream hydrograph and, as a consequence, the calibration and validation phases are not independent. Given the number of degrees of freedom present in such calibration, whereby separate friction parameters can be assigned at each computational node and at each time step, a reasonable correspondence between observed reach outflow data and a calibrated flood routing model (of any dimensionality or spatial resolution) is relatively easy to accomplish. Moreover, there is a strong element of equifinality in this calibration procedure as many different parameterization sets may produce equally acceptable fits to a given set of observed data. The quality of this type of evidence as proof that the model is a robust predictive tool is therefore questionable, and in using such data it has proved impossible to disaggregate error due to model parameterization, structural or discretization errors, data errors or flaws in the calibration procedure itself.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2000
- Accession Number
- ADA381366
Entities
People
- Malcolm G. Anderson
- P. D. Bates
Organizations
- University of Bristol