Author(s): Neil M. Hunter; Paul D. Bates; Matthew S. Horritt
Linked Author(s): Paul Bates
Keywords: Keywords: Flood modelling; Calibration; Validation
Abstract: In this paper we test the performance in prediction of two different formulations of a standard storage cell model for floodplain inundation. Specifically, we compare a model which uses a fixed time step and flux limiter, with a model that uses a time step that adapts in time such that it honours the Courant-Freidrichs-Levy condition for model stability. The models are applied to a 60km reach of the River Severn, UK and used to predict flood wave travel time and inundation extent for two large flood events that occurred in 1998 and 2000 for which observational data were available. Cross calibration-validation tests showed that whilst parameter sets could be identified for both models that simultaneously provided acceptable simulations of flood wave travel time and inundation extent, these occurred over a broader region of the parameter space for the adaptive time step model potentially making it easier to calibrate.
Year: 2005